lexical errors in norwegian intermediate and advanced learners of
TRANSCRIPT
UNIVERSITY OF OSLO
Lexical Errors in Norwegian
Intermediate and Advanced
Learners of English
A thesis presented to
The Department of Literature, Area Studies and European Languages
SPRING 2013
Karina Rose Mahan
i
Abstract ii
Abstract
This thesis builds on previous taxonomies of lexical errors in order to find patterns of lexical
errors in Norwegian intermediate and advanced English texts. This study uses a taxonomy
created by Angela Hasselgren to determine patterns of lexical errors. Lexical errors are
labeled by their routes (how the learner chooses the wrong word), effects (why the word is
wrong) and influences (what causes the error). By locating and labeling lexical errors in two
corpora, this study illustrates how lexical errors are distributed across texts written by
Norwegian speakers of English. The distribution of errors uncovers patterns that in turn
explain how and why some errors are repeated. By contrasting intermediate and advanced
speakers, it is determined that the distributions of intermediate and advanced lexical errors are
significantly different. The amount of errors is significantly smaller in the advanced texts.
Comparative distributions show that intermediate and advanced learners have different ways
of choosing words (routes), but the distribution of effects remains constant. Advanced learners
appear to use direct L1-influence and intralingual influence more, while intermediate learners
depend on indirect L1-influence.
Acknowledgements iii
Acknowledgements
This thesis would never have been written if it weren't for the help and support of
many wonderful people. First I would like to thank Janne Skjelbred, who helped me
compile the intermediate corpus. Without her help, there wouldn't have been an
intermediate corpus to base my thesis on. I would also like to thank Lisbeth Brevik for
her support, encouragement and much-appreciated suggestions.
A special thanks goes to my supervisor, Signe Oksefjell Ebeling. She gave
invaluable advice on many subjects, helped me on very short notice, worked with
extreme deadlines and thoroughly read and corrected this thesis. I would also like to
thank my friends Eric and Aurelija for their technical support on word formatting and
statistical analysis. Additionally I would like to thank my friend Myriam for
thoroughly proofreading my thesis on an extremely short notice.
I found the process of writing a thesis often times demanding, but much less so
in the company of fantastic neighbors, family and friends. Thanks to my neighbors
who always had time for a cup and an ear to lend. I would especially like to thank my
mother and Aurelija for their continual support and patience.
Table of Contents iv
Table of Contents
Abstract......................................................................................................................... ii
Acknowledgements...................................................................................................... iii
Table of Contents......................................................................................................... iv
Figures......................................................................................................................... ix
Tables........................................................................................................................... xi
List of abbreviations..................................................................................................... 1
1. INTRODUCTION.................................................................................................... 3
1.1 Why vocabulary?................................................................................................ 4
1.2 Why lexical errors?............................................................................................. 5
1.3 The current study................................................................................................ 5
2. RESEARCH QUESTIONS....................................................................................... 7
3. WORDS ................................................................................................................. 11
3.1 Words and lexical items.................................................................................... 11
3.2 Word combinations and the relationship between words .................................12
3.2.1 Frequency and word combinations............................................................ 12
3.2.2 The illusion of choice................................................................................ 13
3.2.3 Collocation................................................................................................ 14
3.2.4 Idioms........................................................................................................ 15
3.3 Word knowledge............................................................................................... 15
3.3.1 How does knowledge of words progress? ...............................................16
4. ERRORS AND THEIR SOURCES........................................................................ 19
4.1 Lexical errors.................................................................................................... 19
Table of Contents v
4.2 Sources of errors.............................................................................................. 20
4.2.1 Interlingual influence ................................................................................21
4.2.1.1 Strong L1-influence.............................................................................21
4.2.1.2 Weak L1-influence.............................................................................. 23
4.2.2 Intralingual influence.................................................................................24
4.2.3 Errors – interlingual or intralingual?.........................................................25
5. TAXONOMIES OF LEXICAL ERROR.................................................................27
5.1 Martin's taxonomy of lexical dissonance.......................................................... 28
5.2 A taxonomy of lexical dissonance for Norwegian students of English.............30
5.2.1 Routes........................................................................................................ 31
5.2.1.1 Cognates.............................................................................................. 32
5.2.1.2 Transliterations.................................................................................... 32
5.2.1.3 Perceived Equivalents......................................................................... 33
5.2.1.4 Synonyms............................................................................................ 34
5.2.1.5 Associations........................................................................................ 35
5.2.1.6 Cores and Lexical Teddy Bears........................................................... 36
5.2.1.6.1 What is a core word? ................................................................... 36
5.2.1.6.2 Core words – wrong?................................................................... 36
5.2.1.6.3 Labeling core words..................................................................... 37
5.2.1.6.4 Lexical teddy bears ..................................................................... 38
5.2.2 Effects........................................................................................................39
Table of Contents vi
5.2.2.1 Semantic ............................................................................................. 39
5.2.2.2 Collocational....................................................................................... 40
5.2.2.3 Stylistic/Connotational........................................................................ 40
5.2.2.4 Syntactic.............................................................................................. 40
5.2.2.5 Invalid................................................................................................. 41
5.2.3 Influence.................................................................................................... 41
5.2.4 Shortcomings with the taxonomy.............................................................. 42
6. METHODS AND MATERIAL............................................................................... 45
6.1 Discussion of types of methods and data......................................................... 45
6.1.1 Error Analysis............................................................................................ 45
6.1.2 Translation versus free production – pros and cons...................................46
6.1.4 Cross-sectional data................................................................................... 47
6.2 The data and participants.................................................................................. 48
6.2.1 The intermediate corpus............................................................................ 48
6.2.2 The advanced corpus................................................................................. 48
6.3 Procedure ......................................................................................................... 50
6.3.1 Locating lexical errors............................................................................... 50
6.3.2 Some terms used in the data analysis.........................................................50
6.3.3 Statistical Analysis.....................................................................................51
7. DATA ANALYSIS.................................................................................................. 53
7.1 General Overview............................................................................................. 53
7.1.1 Routes........................................................................................................ 54
Table of Contents vii
7.1.2 Effects........................................................................................................55
7.1.3 Influences.................................................................................................. 56
7.2 Intermediate and Advanced Comparisons ........................................................ 56
7.2.1 Routes........................................................................................................ 57
7.2.1.1 Comparison of tokens..........................................................................58
7.2.1.2 Comparison of types............................................................................59
7.2.1.3 The repetition-ratio ............................................................................. 60
7.2.1.4 Statistical significance......................................................................... 61
7.2.1.5 Comparative distributions................................................................... 62
7.2.1.6 Summary of routes.............................................................................. 63
7.2.2 Effects........................................................................................................64
7.2.2.1 Comparison of tokens..........................................................................64
7.2.2.2 Comparison of types............................................................................65
7.2.2.3 The repetition-ratio.............................................................................. 66
7.2.2.4 Statistical significance......................................................................... 66
7.2.2.5 Comparative distribution .................................................................... 67
7.2.2.6 Summary of effects............................................................................. 68
7.2.3 The route/effect relationship...................................................................... 69
7.2.3.1 Distribution of effects in routes........................................................... 70
7.2.3.2 Summary of effects in routes...............................................................73
7.2.4 Influence.................................................................................................... 74
Table of Contents viii
7.2.4.1 Routes................................................................................................. 74
7.2.4.1.1 Distribution of inter- and intralingual influence in L2-based routes
.................................................................................................................... 75
7.2.4.1.2 Statistical significance.................................................................76
7.2.4.1.3 Comparative distribution..............................................................77
7.2.4.2 Effects................................................................................................. 77
7.2.4.1.2 Distribution of influences in effects.............................................78
7.2.4.3 Summary of influences........................................................................81
8. DISCUSSION......................................................................................................... 83
8.1. The first research question............................................................................... 83
8.2 The second research question........................................................................... 90
9. CONCLUSION....................................................................................................... 93
9.1 Pedagogical implications.................................................................................. 94
9.2 Suggestions for further research....................................................................... 97
9.3 Final words....................................................................................................... 99
10. REFERENCE LIST............................................................................................ 100
APPENDIX 1............................................................................................................ 104
APPENDIX 2............................................................................................................ 108
APPENDIX 3 (Intermediate Dissonances)................................................................ 110
APPENDIX 4 (Advanced Dissonances)....................................................................124
APPENDIX 5............................................................................................................ 136
Figures ix
Figures
List of Figures
Figure 1: Progress of aspects of word knowledge in the early stage............................16
Figure 2: Progress of aspects of word knowledge in the developing stages.................17
Figure 3: Progress of aspects of word knowledge in the advanced stages....................17
Figure 4: Division of semantic space for English believe/think and Swedish tro/tycka/tänka (Carter 2012: 20)................................................................................. 22
Figure 5: Distribution of routes in corpora................................................................... 53
Figure 6: Distribution of effects in corpora.................................................................. 53
Figure 7: Percentage distribution of routes and effects in corpora...............................54
Figure 8: Distribution of influence in corpora..............................................................56
Figure 9: Ranking of routes in tokens for intermediate and advanced students...........58
Figure 10: Ranking of routes in tokens for intermediate and advanced students.........59
Figure 11: Repetition-ratio for routes in intermediate and advanced students..............60
Figure 12: Comparative distribution of routes in intermediate students.......................62
Figure 13: Comparative distribution of routes in intermediate students......................62
Figure 14: Comparative distribution of routes for intermediate and advanced students..................................................................................................................................... 62
Figure 15: Ranking of effects in tokens for intermediate and advanced students........64
Figure 16: Ranking of effects in types for intermediate and advanced students..........65
Figure 17: Repetition-ratio of effects for intermediate and advanced students...........66
Figure 18: Comparative distribution of effects for intermediate students...................67
Figure 19: Comparative distribution of effects for advanced students........................67
Figure 20: Comparative distribution of effects for intermediate advanced students....68
Figure 21: Distribution of effects in routes for intermediate students.........................70
Figure 22: Distribution of effects in routes for advanced students..............................70
Figures x
Figure 23: Distribution of effects in synonyms and cores for intermediate students...72
Figure 24: Distribution of effects in synonyms and cores for advanced students........72
Figure 25: Distribution of influence in L2-based routes for intermediate students.....75
Figure 26: Distribution of influence in L2-based routes for intermediate students.....75
Figure 27: Comparative distribution of influence in intermediate students.................77
Figure 28: Comparative distribution of influence in advanced students......................77
.................................................................................................................................... 77
Figure 29: Distribution of influence in effects for intermediate students...................78
Figure 30: Distribution of influence in effects for advanced students........................78
Tables xi
Tables
List of tables
Table 1. Comparable information about the two corpora used..................................... 49
Table 2. LL-values to p-values..................................................................................... 51
Table 3. Number of total dissonances found in the Norwegian learner data (intermediate and advanced)........................................................................................ 53
Table 4. Number of total dissonances found in intermediate versus advanced corpora56
Table 5. Comparable information of dissonances in the intermediate and advanced corpus........................................................................................................................... 57
Table 6. Total number of tokens and types in routes for intermediate and advanced corpora ........................................................................................................................ 58
Table 7. LL-values and p-values for statistical differences between intermediate and advanced routes............................................................................................................61
Table 8. Total number of tokens and types in effects for intermediate versus advanced students........................................................................................................................ 64
Table 9. LL-values and p-values for statistical differences between intermediate and advanced effects........................................................................................................... 67
Table 10. The route/effect combination of all dissonances for intermediate students. .69
Table 11. The route/effect combination of all dissonances for advanced students.......69
Table 12. Distribution of influences in intermediate and advanced dissonances..........74
Table 13. Inter- and intralingual influence in the L2-based routes...............................75
Table 14. LL-values and p-values for statistical differences between intermediate and advanced influences..................................................................................................... 76
Table 15. Total number of influences found in effects for intermediate students......78
Table 16. Total number of influences found in effects for advanced students............78
1
List of abbreviations
EFL – English as a Foreign Language
ICLE – International Corpus of Learner English
L1 – first language
L2 – second language
NICLE – the Norwegian part of the International Corpus of Learner English
PE – perceived equivalent
SLA – second language acquisition
2
3
1. INTRODUCTION
Language is traditionally divided into grammar and lexis. In second language
acquisition/learning, they take the form of rule-learning (grammar) and vocabulary learning
(lexis). Both are important to be able to communicate, yet second language research focuses
much more on grammar. There is extensive research done on how learners develop
grammatical patterns, what mistakes are typical of learners and how rules can be applied to
prevent grammatical errors. Perhaps as a result of this, second language teaching puts much
more emphasis on grammar than vocabulary in the second language classroom. This is done
by explicitly teaching grammatical rules, and assuming that students will acquire vocabulary
through osmosis. Most foreign language teachers have therefore been taught how to cope with
grammatical errors (They is* is a concord mistake), but not how to cope with lexical errors
(We had a jolly nice time is a wrong word choice – but what makes it wrong?).
Underdeveloped systems for vocabulary have large consequences for the EFL (English as a
Foreign Language) classroom – teachers are left telling their student which word is right (we
had a jolly good time), without offering a satisfactory explanation for why the other is wrong.
Being able to explain why something is wrong is the first step to preventing similar errors. If
systems for vocabulary are not developed, these mistakes can only be treated as singular
errors that simply must be memorized.
Lexis has largely been ignored in SLA (second language acquisition) studies until the
last 20-30 years. Researchers have blamed the “unstable and unsystematic nature of the
lexicon” to justify their focus on grammatical aspects (Llach 2011: 70). To many, the lexicon
is ostensibly a collection of unrelated, lone words that cannot be systematized. However,
quantitative research on word frequencies can now confirm that words rarely occur alone in
discourse. Psycholinguistics is also pointing toward a highly organized mental lexicon, where
words are organized by both semantic (meaning-based) and formal aspects (such as similar
word forms) (Aitchison 2012: 10). Such research suggests that a) words can be grouped
together based on how they occur in discourse and b) words can be grouped together based on
how they are acquired.
In SLA studies, techniques for identifying, classifying and systematically interpreting
errors (also known as error analysis) are constantly being developed to account for the many
errors that second language learners make (Crystal 2008: 173). Analysis of lexical errors is the
first step to creating taxonomies of errors – i.e., systems that can explain how/why words are
wrong. Quantitative studies of these mistakes in learner language can also determine which
4
mistakes are more reoccurring and characteristic of non-native speakers. This thesis is such a
study. By examining texts written by Norwegian students of intermediate and advanced
English, this study hopes to establish a) what the most reoccurring lexical mistakes in
Norwegian speakers of English are and b) whether these reoccurring lexical mistakes are the
same in both intermediate and advanced students.
1.1 Why vocabulary?
Vocabulary is the largest area of linguistic knowledge (Caspi & Lowie 2010: 46). Adult native
English speakers with an educated background know around 15, 000 – 20, 000 word families
(words with the same base, but belonging to different word classes, e.g. beauty - beautiful)
(Schmitt 2010: 29). An L2 (second language) speaker must acquire approximately 3, 000
word families to minimally follow a conversation, and around 9, 000 word families to read
novels or newspapers (Ortega 2009: 88). Unsurprisingly, lack of vocabulary does not only
impede knowledge of the language – Meara comments that measurements of vocabulary size
have been shown to correlate positively with proficiency levels in reading and writing, and in
general language proficiency (Meara 2009: 34).
Vocabulary knowledge is two-sided – it is not only about how many words you know
(vocabulary breadth), but also how well you know them (vocabulary depth). Studies in
vocabulary acquisition have shown that learning vocabulary is not an instantaneous process –
words are not either known/unknown, and knowledge of words most likely operates on a
continuum (see section 3.3.1). For instance, Meara discovered that changes in response to
words (e.g. moving from clang associations to native-speaker-like responses) still take place
as much as twelve weeks after initial presentation of taught words (Meara 2009: 27). He also
believes that knowing words is a “complex and multi-faceted skill, perhaps best described in
behavioral terms as the ability to react to a word in ways which are considered appropriate by
the speech community” (Meara 2009: 19). Many lexical researchers have noted that word
knowledge works on several levels, from explicit levels (such as form-meaning linkage) to
implicit levels (such as when it is appropriate to use a word and around whom). Word
knowledge can also be receptive (speaker is only able to recognize a word) and productive
(speakers can also produce word). Productive knowledge typically lags behind receptive
knowledge by a few hundred words, and once enough knowledge of a word is acquired, it
changes from being passive to active knowledge (Meara 2009: 30).
Perhaps due to a) the large number of words that need to be acquired to communicate
5
and b) the many levels and aspects of knowing a word, it has been shown that L2 students
think a lack of vocabulary is one of their major problems in their production of English
(Simensen 1998: 220).
1.2 Why lexical errors?
The L2 vocabulary, much like L2 grammar, is often riddled with mysterious mistakes that a
native speaker would not make. Finding systems for these errors may explain how and why
these mistakes occur. Error analysis researchers hope that by investigating these special L2
word errors, we might gain insight into how L2 speakers acquire words and why they use
them in non-native ways. This is called the study of lexical errors, and examines how L2
speakers misuse or misconceive vocabulary. The importance of lexical errors can be
highlighted by two findings:
1. Researchers have suggested that lexical errors are as much as three times more
frequent than grammatical errors in second language learners (Blaas 1982, cited in Ellis 2008:
50).
2. Research has shown lexical errors cause more problems of interpretation and a higher
degree of irritation to the native speaker (Johansson 1978: 71).
The problem with locating lexical errors is that researchers cannot agree on what lexical
knowledge consists of. If we cannot establish what lexical knowledge is, how can we find
systems to judge if someone has misunderstood the knowledge? Erroneously, many believe
that since there are “so many words” in a language, it is impossible to say why picking a
certain one will be wrong. Studies of lexical knowledge and errors can tell us that although
there are (potentially) an infinite number of words and word combinations a speaker has at
her/his disposal, the type of information connected to a word is finite (for instance, spelling,
pronunciation, meaning, collocation and frequency), and therefore incorrect use of this
information is also finite (an error of spelling, pronunciation, meaning, et cetera).
1.3 The current study
By applying methods from SLA (specifically error analysis) and corpus linguistics, and
drawing on tenets of semantics and phraseology, this thesis will attempt to uncover
distributional patterns of lexical errors in intermediate and advanced texts produced by
Norwegian speakers of English. A basis for this study was presented in chapter 1. Chapter 2
6
provides the research questions which this thesis will answer. Llach points out that one can
only find lexical errors based on a) what “knowing a word” implies and b) what counts as a
word (Llach 2011: 72). Therefore chapter 3 is dedicated to clarify what a word is and what
“knowing a word” entails. Chapter 4 will explain lexical errors and their sources, and chapter
5 will describe the taxonomy of lexical errors in this study. Chapter 6 consists of methods and
materials. Chapter 7 is the data analysis, and chapter 8 will answer the research questions
based on said analysis. Chapter 9 will conclude the main findings, suggest applicable uses for
the findings in the EFL classroom and provide suggestions for further research. The data
which the thesis bases itself on (the number and types of errors) can be found in full detail in
appendices 3 (intermediate learner errors) and 4 (advanced learner errors).
7
2. RESEARCH QUESTIONS
This research paper is inspired by Angela Hasselgren's MA thesis (hovedoppgave) “Right
Words, Wrong Words and Different Words: an investigation into the lexical coping of
Norwegian advanced learners of English” (Hasselgren 1993). Hasselgren investigates how
Norwegian university students cope with their L2 (English) lexicon by a) analyzing how the
group selects wrong words in translation tasks, b) determining why the selected words are
wrong and c) possible sources of these errors. The wrong words are lexically dissonant1
because they create dissonance between the word and its context, rather than violating any
rules of the language (Martin 1984: 130). Hasselgren develops a taxonomy for labeling
different aspects of lexical dissonance in hopes that the taxonomy could explain all such
dissonances based on certain, limited criteria (e.g. that there are five general ways a word can
be wrong). Quantitative research of dissonances based on the taxonomy could aid language
teaching by identifying what aspects of vocabulary L2 speakers struggle with the most, and
diagnose exactly why an L2 vocabulary – even a grammatically flawless advanced learner's
vocabulary – may strike a native speaker as “foreign” (Hasselgren 1993: 2). The lexical
dissonances in Hasselgren's study are labeled by their route (why they are wrong), effect
(how they are wrong) and divergence (the source of error, i.e. inter- and/or intralingual
influence) (see section 5.2 for more information on routes, effects and divergence).
Hasselgren investigates dissonances in translation texts2 written by Norwegian
speakers of advanced English. The focus of this study is to see how lexical dissonance is
manifested in intermediate speakers as well as advanced. Such a comparison may provide
insight into if there are any differences in routes, effects and divergence between the
intermediate and the advanced stage in Norwegian learners of English. By analyzing free
production texts written by intermediate and advanced students, this study builds on or
complements Hasselgren's study of lexical dissonance in Norwegian speakers of English.
1 The words dissonance and dissonances are used interchangeably in this thesis, based on whether it is one mistake (a dissonance) or if it refers to lexical errors in general (dissonance). Martin (1984) prefers the term dissonances to explain individual word errors (.e.g one dissonance), whereas Hasselgren (1993) uses dissonance as an uncountable term for the phenomenon, and uses wrong word in the case of countable.
2 Tasks where the L2 speaker translates from their L1 (Norwegian) to their L2 (English)
8
The main research question for this study is:
1. How is lexical dissonance manifested in free production texts written by Norwegian
speakers of English (intermediate and advanced)?
A taxonomy for labeling lexical dissonance is provided, including routes, effects and
influence. The main research question can thus be answered by several subquestions:
a) How is lexical dissonance distributed across routes and effects in Norwegian
intermediate and advanced learners of English?
b) If some routes and effects are larger than others, what are possible reasons for
this?
Each dissonance is assigned both a route (how it is wrong) and effect (why it is wrong). One
subquestion of interest is therefore:
c) How do routes and effects combine? Is there a discernible pattern between how a
wrong word is chosen and why it is dissonant?
These questions do not fully cover the question of influence, or source of dissonance. One
aspect that should be examined is how large a role the first language plays in second language
error production. There are several views on L1-influence on the target language. Ellis claims
that “a large number – and in some cases perhaps most – of the errors that learners produce
are intralingual in origin rather than transfer” (Ellis 2008: 55). A subquestion that can confirm
or refute this statement is:
d) What is the proportion of L1-influenced dissonances (interlingual) vs non-
influenced L1-dissonances (intralingual)?
Hasselgren's study has the following route distribution ranking: synonyms > transliterations >
associations > cognates = perceived equivalents. Effects have the following distribution
ranking: semantic > collocational > stylistic/connotational > invalid > syntactic. Since this
study uses her taxonomy, it would be interesting to see if the categories hold the same rank in
my data. Therefore, the last subquestion for the first research question is:
e) Will my data have the same distributional patterns for routes and effects as
Hasselgren's?3
3 This is determined by a token ranking (see sections 6.3.2 and 7.1.1 for more information)
9
Since this study will analyze lexical dissonance at both the intermediate and advanced levels
of Norwegian speakers of English, it is necessary to take into account that different levels of
English may produce different distributions of dissonance, e.g. that intermediate students have
different ways of choosing the wrong word than advanced students. The routes, effects and
influence distributions are of course useful to look at from a wider perspective, but it is also
necessary to investigate if there are differences between the intermediate and advanced
learners. To take this factor into account, a second main research question is therefore posed:
2. Will the number and distribution of lexical dissonances differ between the
intermediate and advanced students?
This research question can also be split into several, more specific research questions,
including:
a) Is there a difference in number of lexical dissonances found between intermediate
and advanced students?
b) Is there a difference in distribution of lexical dissonance between routes?
c) Is there a difference in distribution of lexical dissonance between effects?
d) If there are differences between intermediate and advanced students, are these
differences statistically significant?
Taylor (1975) states that “with increased proficiency in the target language, [L2 speakers] rely
proportionately less frequently on their native language grammar, and rely more frequently on
their ever-increasing knowledge of the target language” (Taylor 1975: 88). This statement
inspired a subquestion regarding L1-influence:
e) Will the advanced students show proportionally less L1-influence in lexical
dissonance than the intermediate students, illustrating “reliance on their ever-increasing
knowledge of the target language” as suggested by Taylor (1975)?
10
11
3. WORDS
Llach states that defining a lexical error is only possible once a word is defined (Llach 2011:
72). As words are not always as clearly defined as we think or wish, this section will clarify
what is understood by the terms “word” and “lexical item,” examine other common ways of
classifying words, and look at some aspects of word knowledge.
3.1 Words and lexical items
There are several ways of classifying what constitutes a word. A word can be classified in
terms of levels, such as the written, phonological and syntactic level (Saeed 2009: 56). Words
can also be classified in terms of theoretical distinctions. Crystal suggests that there are three
common theoretical distinctions of words, where words are defined based on if they are a)
physically definable units divided by space, b) lexemes (variants of the same unit) or c)
grammatical words (Crystal 2008: 522).
For this thesis it is more practical to define a word (and thus a lexical error) based on
meaning rather than orthographic, syntactic or other features. For instance, the dissonant
phrase: As late as yesterday (as recently as yesterday) illustrates that the student has chosen
one wrong word, but it is only wrong because it does not fit in with the rest of the phrase. In
other words, for a word to be wrong, it often has to be incompatible with the surrounding
words, and these surrounding words can cluster together to form one unit of meaning.
Cruse explains that words can range from being the smallest mobile units in a
sentence, to the largest units that “resist interruption” (Cruse 1986: 35-36). An uninterrupted
unit can stretch from the single word recently to the phrase as recently as [time specification].
However, not many native English speakers would agree that as recently as is a single word.
This is where the term “lexical item” comes in. Lexical items are generally viewed as one or
several items which convey one meaning. This thesis counts such lexical items as one word or
one dissonance – a “wrong word” can mean anything from a single word to an uninterrupted
stretch of meaning which consists of several words.
Moreover, a common distinction is that of lexical words and grammatical words.
Lexical words comprise main verbs, nouns, adjectives and adverbs, while grammar words
consist mainly of determiners, pronouns, auxiliary verbs, prepositions and conjunctions
(Hasselgård et al 2007: 14). As the focus of this study is on lexical errors, only lexical words
will be analyzed.
12
Another common way of classifying words is if they belong to the same word family.
A word family is a group of words that are grouped together on the basis of their morphology.
They have a base form, possible inflectional forms and words that derive from them by
prefixation and suffixation (Jackson & Amvela 2007: 19). For example, the word family
PERFECT consists of the noun perfection, the verb to perfect (and all its subsequent verb
forms), the adjectives perfect and imperfect and the adverbs perfectly and imperfectly. The
term “word family” is useful for vocabulary acquisition researchers, because although a
person may “know” 20, 000 words in a language, s/he may in reality only know, say, 5, 000
word families. This study uses the concept of word families to account for dissonance where
the word base is correct (i.e. that it belongs to the right family, such as perfect), but the word
form is incorrect (e.g. a perfectly day).
3.2 Word combinations and the relationship between words
Words do not exist in isolation. Their meanings are defined through their relationships with other words and it is through understanding these relationships that we arrive at our understanding of words (Richards 1976: 81).
This thesis has mentioned “units of meaning,” but not what these units of meaning consist of.
For instance, in the case of as late as yesterday, it was determined that late is a wrong word
choice. What constitutes these choices? This section will discuss “units of meaning” and how
they relate to lexis.
3.2.1 Frequency and word combinations
One cannot talk about units of meaning, or indeed lexical choice, without discussing
phraseology. Phraseology is the study of the structure, meaning and use of word combinations
(Cowie 1994, quoted in Granger & Paquot: 27). Phraseology stresses that words are not lone
units, and there are often patterns of co-occurrences – i.e. some words combine more often
than others. Frequency is key to identifying word combinations – the more frequent two
words co-occur (blond hair, but not blond car*), the stronger the relationship between two
words are said to be. Researchers believe that a speaker's ability to map and determine
frequency and co-occurrence of words can only come through vast amounts of exposure to the
13
language (Schmitt 2010: 31). However, if words are taught in context (e.g. information about
frequency of a word and its possible combinations are provided), it is believed that learners
can use this information to quickly sound more native-like without being exposed to the
language over a long period of time.
3.2.2 The illusion of choice
Sinclair, who is one of the advocates of phraseology, states that the role of words in sentences
can be viewed in two ways: through the open-choice principle (few constraints on
sentence/phrase structures other than grammatical ones which can be explained) and the
idiom principle (semi-constructed phrases that constitute single choices, even though they
might appear to be analyzable in segments) (Sinclair 1991: 110). The open-choice principle is
problematic, because although there should be limitless ways of combining words that do not
violate the rules of syntax, native speakers “do not exercise the creative potential of syntactic
rules to anything like their full extent” (Pawley & Syder 1983: 193). This is referred to as the
puzzle of native-like selection – how native speakers can distinguish between “which well-
formed sentences are native-like” (Pawley & Syder 1983: 194).
The idiom principle can explain these native-like choices, by stating that these choices
are determined outside the rules of grammar. There are many terms for these semi-constructed
phrases including “formulaic language,” “prefabricated language” and “phraseological units,”
to mention a few. The strength of how set this prefabricated language is can vary from words
that are loosely grouped together, such as open slots (a good/great/bad/horrible time), to
idioms that cannot substitute a single word (all bark and no chomp*).
Prefabricated language is no small part of language. A study conducted by Erman and
Warren suggests that over 50% of a text will consist of prefabricated language, and for 100
words, only 71 word choices are possible (Erman & Warren 2000: 50). Aitchison postulates
that multi-word expressions outnumber single words by ten to one (Aitchison 2012: 115).
These types of prefabricated language are so vital to how speakers perceive lexis that there is
a growing body of work indicating they are stored as automatized units in memory (i.e. that
they are memorized and perceived as one “chunk”) (Wood 2001: 579). The speed of retrieval
of prefabricated language is so fast that this suggests prefabricated language is stored as a
single unit in the mind. For native English speakers, the retrieval and combining of lexical
chunks and frames is the foundation of fluency (Wood 2001: 582).
Although a speaker has hundreds of thousands of words at their disposal, the
14
restrictions on these words are immense and largely a result of co-occurrence frequencies. The
ways in which words can combine can be utterly confusing to the non-native speaker, since
there are no perceivable “rules” other than that some words tend to group more frequently
with others. Followers of Sinclair and the idiom principle will stress the limitations of choice
in lexis, and that these limitations are so imbedded in the language that they even affect how
native speakers perceive and process language. L2 speakers sound “foreign” because they
“exceed” these limitations.
3.2.3 Collocation
Strongly connected to the idea of the idiom principle is the concept of collocation, which can
account for a large amount of lexical dissonance. Collocation is the frequent and habitual co-
occurence of words in a language (Fiedler 2007: 189). It is divided into two types: lexical
collocation (two lexical items: strong tea) and grammatical collocation (lexical word and
grammatical word: bring up). Collocation consist of a node word (tea) and its collocate
(strong). The node word and collocate are so inter-dependent that if two words which cannot
collocate with each other are combined, dissonance is created between the node word and its
collocate (muscular/sturdy/tough tea*). For the purpose of this thesis, lexical collocation is
viewed as one lexical error. For instance, sturdy tea can be said to be one lexical item, and
therefore will only be counted as one error.
Grammatical collocation consists mainly of multi-word verbs and delexical verbs.
Multi-word verbs are verbs where there is a main verb and one or two particles that are
viewed as constituents of a single unit (Jackson & Amvela 2007: 75). Some multi-word
combinations are: prepositional verbs (apply for, laugh at), phrasal verbs (bring up, count
[someone] in), phrasal-prepositional verbs (look up to, catch up on) and combinations that
include three or more constituents (make do with, take care of) (Hasselgård et al 2007: 154-
155). Delexical verbs are (main) verbs used so frequently that the meaning of the verb
becomes reduced and is more dependent on the patterning of the words around it (Hasselgren
1993: 26). Hasselgård et al. list the five most common delexical verbs in English: give, have,
make, take and do. Examples include: take a look around, have a chat (Hasselgård et al 2007:
156-7). Prepositions are disregarded in this study, as they are grammatical words, so
prepositions in lexical items such as in bring in a topic are not counted. Only the lexical word
can be counted in grammatical collocation (e.g. take up a discussion, instead of bring up).
Collocation links words together independent of sense. Aitchison posits: “Word
15
meaning is probably learned by noting the words which come alongside” (Aitchison 2012:
114). In other words, although there is no direct sense link between collocates, a reason for
why there are such strong links between seemingly unrelated words could be because we
create meaning between words based on how they are grouped together, independent of
semantic fields. Lexical dissonance is termed “dissonance” because a large number of lexical
errors are due to the fact that non-native speakers cannot see the link between these non-
sense-related words – the link is only apparent through repeated exposure to the word
combinations.
3.2.4 Idioms
Idioms are also regarded as one unit of meaning in this study of lexical error. There are
several definitions of idioms. Jackson and Amvela explain that they can be viewed as: “a type
of collocation involving two or more words in context [or] a type of multi-word lexeme,” but
that in general terms, an idiom is “defined as a phrase, the meaning of which cannot be
predicted from the individual meanings of the morpheme it comprises” (Jackson & Amvela
2007: 77). This study counts both definitions as being a “lexical unit.” It was found necessary
to include them because wrong word choices can occur in large stretches of units of meaning.
For instance, the phrase want nothing to do with is marked as stylistically dissonant in one of
the student texts. It is impractical to label this as 3 lexical errors (want, nothing and do), since
it is clearly the entire unit of meaning that has been misunderstood. Therefore, idioms, or
longer stretches of units of meaning, have to be taken into account in this analysis of lexical
errors.
3.3 Word knowledge
What does it mean to “know” a word? Or, how much information can a speaker deduce from
a word? These are questions that many researchers have attempted to answer, particularly in
relation to second language acquisition and learning, where learning words is a much more
active process.
Jack Richards (1976) is the first to explicitly express that each word has several
aspects of knowledge, and only by knowing these aspects, can we fully “know” a word. He
lists what he believes these aspects of word knowledge are, with the purpose of offering a
frame of reference for assessing vocabulary teaching. He examines linguistic,
psycholinguistic, and sociolinguistic aspects of word knowledge in his article, including word
16
frequency, vocabulary growth in native speakers, collocation, register restraints, case
relations, underlying forms, word association, and semantic structure (Richards 1976: 77). He
famously formulates them as seven word knowledge assumptions, and these are regarded as
the basis of the mapping of word knowledge (Richards 1976: 79-83).
Paul Nation, one of the biggest names in vocabulary acquisition research, refines and
builds on Richards' word knowledge assumptions, listing the following aspects of word
knowledge: meaning, written form, spoken form, grammatical characteristics,
collocation, register constraints, frequency and associations (Nation 1990, quoted in
Schmitt 2010: 31). Nation also includes other features such as orthographic and phonological,
which are perhaps the most common measurements when people assess their own knowledge
of a word (as in, “can I spell and pronounce the word?”).
3.3.1 How does knowledge of words progress?
The acquisition of some aspects of word knowledge is not straightforward. Contextually-
dependent aspects of word knowledge, such as collocation, meaning and associations can
contain a lot of information. For instance, one word can have several meanings depending on
context, and the more meanings a speaker is aware of, the more knowledge the speaker has of
the word. Henriksen (1999) suggests that word knowledge is not a question of known or
unknown – it is a continuum from zero to partial to precise knowledge of a word. Schmitt,
using Nation's suggested eight aspects of word knowledge, creates a continuum for each
aspect, proposing how learners of English progress along this continuum (Schmitt 2010: 37):
Figure 1: Progress of aspects of word knowledge in the early stage
Figure 2: Progress of aspects of word knowledge in the developing stages
17
Figure 3: Progress of aspects of word knowledge in the advanced stages
In Schmitt's continuum of word knowledge, spoken form, written form and grammar are
mastered first, perhaps indicating that these types of knowledge are limited and can be more
easily mastered. Collocation, register, frequency and associations lag behind at all stages.
Schmitt explains that these four categories are much more difficult to acquire because they do
not occur in isolation and can therefore only be acquired through large amounts of exposure to
English (Schmitt 2010: 31).
This continuum of word knowledge is relevant to this thesis because it examines
“incorrect word hypotheses” (labeled “effects.” See section 5.2.2). An incorrect hypothesis is
the way a learner assumes something incorrectly about a word. These incorrect hypotheses
can be incorrect at the level of “meaning,” (labeled semantic) “collocation,” “register,”
(labeled style), “grammar” (labeled syntax). They occur mainly when the learner only has
partial knowledge of a word (i.e. early or developing knowledge of a word). An abundance of
incorrect hypotheses in one aspect of word knowledge (for instance noticeably more
collocational errors than other types of error) may indicate that the particular aspect is more
difficult to acquire. This will not be examined in this study, but is something to keep in mind
during the data analysis.
18
19
4. ERRORS AND THEIR SOURCES
An error is an unacceptable form of language found in spontaneous speaking or writing
(Crystal 2008: 173). Errors occur if a speaker only has partial knowledge of a language, and
are assumed to reflect a speaker's level of competence in a systematic way (Crystal 2008:
173). Error is defined by (un)intention – an error “only arises when there is no intention to
commit one” (James 1998: 77). If a linguistic element is used in a different (perceivably
wrong) way intentionally, it is not erroneous, but deviant. This is problematic for corpus data,
because errors can only be deduced through context – it is unknown whether a learner
intentionally deviates or not.
There are several degrees of error. James distinguishes between slips (lapses that are
quickly detected), mistakes (can only be corrected if pointed out by someone else), errors
(cannot be self-corrected until input has been provided) and solecisms (breaches of rules that
even native speakers will make, e.g. split infinitives) (James 1998: 83). Mistakes and errors
are of primary interest in error analysis, since slips do not necessarily reflect a
misunderstanding in word knowledge, and solecisms are not a distinct L2 trait. Corpus data
cannot easily distinguish between mistakes and errors, because there is no insight into the
learner's mind. Therefore, all unacceptable, non-native forms (slips, mistakes and errors) can
only be labeled as “errors.”
This section will explain what is meant by “lexical error” and examine some of the
sources of errors.
4.1 Lexical errors
Defining exactly what a lexical error is is problematic and complex. Generally “lexical error”
refers to “the deviations in the learner's production of the L2 norm with regards to the use in
production and reception of lexical items” (Llach 2011: 71). The line between grammar and
lexis is fuzzy, so agreeing on what a lexical error is in relation to a grammatical error is is not
an easy task. Some linguists interpret “lexical error” as meaning “all errors that are not
grammatically fit” (Llach 2011: 73). Others view “lexical error” as a superordinate term for
classes of errors such as word formation, collocation, form/semantic confusion and wrong
word choice (Llach 2011: 73). Because of disagreement on what lexical errors entail, a large
range of words is used to describe lexical error based on different definitions of what lexical
errors represent. Examples include: “wrong lexical choice,” “lexical deviances,” “vocabulary
20
errors,” “incongruencies in lexical gridding,” “semantic deviation” “lexical confusions” and
“lexical simplification,” to mention a few (Llach 2011: 74). Despite the fact that researchers
are reluctant to agree on exactly what lexical errors are and how they can be classified, there
is a general consensus that lexical errors are a response to systematic causes and can be
explained, classified and generalized (Llach 2011: 74). Research on lexical errors is often
devoted to defining what a lexical error is to the particular study, creating a comprehensive
taxonomy and then unearthing the systematic causes of lexical error for the data in question.
As Llach puts, it “particular lexical error types are named depending on the dimension of the
lexical error that prevails in the research,” i.e. the definition and role of lexical errors is
dependent on the data, and often taxonomies are built based on these needs (Llach 2011: 74).
In this study, a lexical error is a lexical item that deviates from L1-use. The lexical
errors are referred to as “dissonances.” One can say that “lexical error” and “dissonance” can
be used interchangeably, but the term “dissonance” is preferred in this thesis. The term “error”
“carries overtones of downright wrongness of meaning,” hence the lengths researchers will go
to to find other terms to describe this phenomenon (Hasselgren 1994: 238). A failure in form-
meaning linkage (e.g. thinking that the word cup represents a vase) is perhaps the only way a
learner's perception of a word can be downright wrong. This is because the speaker has failed
to understand either the reference of the word (what a cup represents in the real world) or the
form of the word (that it is called a vase). Most other instances of lexical errors are caused by
a wrongly chosen word creating “dissonance between the word and its context” (Martin 1984:
130). There is often nothing wrong with the actual sense or understanding of individual words
in “wrong word choices,” such as the collocational mistake nerves of iron. Context is key in
detecting lexical errors – even instances of form-meaning linkage can only be located through
mismatch in context in free production texts, e.g. She dropped her phone on the ceiling
instead of floor.
4.2 Sources of errors
Richards and several other linguists have observed that non-native speaker errors derive from
two major sources: interlingual and intralingual influence. In my taxonomy of lexical errors,
source of error is one of the aspects which will be investigated. In this section, an account of
these two sources of errors will be given.
21
4.2.1 Interlingual influence
Interlingual influence is influence which occurs between two or more languages. In
practice, this normally means from a speaker's first language to their second or third language
(or however many languages a speaker learns after their native language), but in many
instances this is not the case. Observations from language acquisition studies have shown that
all previously learned languages can influence additional language learning (i.e., L1 and L2
can both influence the L3 by facilitating or hindering L3 learning), and evidence is growing
for cases of bidirectional transfer, in which influence can work both ways, from L1-L2 and
L2-L1 (Ortega 2009: 48; 50). Therefore, many linguists prefer the term “cross-linguistic
influence,” because it is:
theory-neutral, allowing one to subsume under one heading such phenomena as transfer, interference, avoidance, borrowing and L2- related aspects of language loss and thus permitting discussion of the similarities and differences between these phenomena ( Sharwood & Kellerman 1986: 1, as cited in Ellis: 351).
This thesis uses the terms “L1-influence” and “interlingual influence.” This is because
although there is of course a strong case for cross-linguistic influence, the participants in this
study only speak two languages (English and Norwegian), and only one type of interlingual
influence will be examined – the influence from their L1 (Norwegian) to their L2 (English).
There are many ways of identifying L1-influence in L2 errors. This thesis
differentiates between two types of influence: strong influence and weak influence. In lexical
studies, strong influence occurs when an L2 speaker chooses a word or structure that is wrong
in their L2, but right in their L1 (e.g. he is high instead of tall (høy)). The assumption is that
the L2 speaker makes the lexical choice because of interference from their first language. This
study also takes weak L1-influence into account, which for this thesis means words that are
wrongly chosen because of the way some words behave differently in the L1 and L2. It is not
a particular L1 word that is the cause of error, but rather that for example Norwegians who
choose the word human instead of person, as in she's a nice human, will have an interference,
because Norwegian only has one superordinate word for the two English words (menneske).
Sections 4.2.1.1 and 4.2.1.2 will explain some of the strong and weak L1-influences in detail.
4.2.1.1 Strong L1-influence
When influence from one language causes a speaker to create a similar structure in the other
language, it is often referred to as “transfer,” i.e. that the speaker transfers their knowledge
from one language to the other. Transfer is an important part of second language vocabulary
22
acquisition, as “generally the cross-linguistic influence between non-native languages in a
European context has been shown to occur primarily in lexis” (Ringbom 1987: 114). There
are two major types of lexical transfer from the L1 to the L2 that will be examined in this
study: features of an L1 item that are assigned to an L2 item (lexical transfer), and L2 items
which are combined in compounds or phrases analogously with the L1 structure (loan
transfer) (Ringbom 1983: 207, quoted in Hasselgren 1993: 13)
Semantic lexical transfer is the extension of a word in the L1 to cover the semantic
space of more than one L2 item. This can be explained in part by the semantic equivalence
hypothesis. When two words are approximately equivalent to each other in another language,
such as think and tro, Ijaz (1986) postulates that learners will rely on a “semantic equivalence
hypothesis” (cited in Hasselgren 1993: 15). The learner will assume the semantic boundaries
of an L2 item are identical to the L1 equivalent. Words cover different semantic spaces, and
although a word in the L1 and the L2 may be translatable on some level, often the semantic
spaces these two words cover will differ between the two languages. Take think and tro
(Swedish), which have approximately the same main sense and can often be used in the same
phrases and collocates in both English and Swedish. Then see how the semantic space is
divided for the two:
Figure 4: Division of semantic space for English believe/think and Swedish tro/tycka/tänka (Carter 2012: 20)
A Swedish speaker of English who tries to directly translate tro to think may discover that
her/his use of think is often wrong, since tro can also mean a personal belief or opinion,
whereas think cannot. Conversely, English speakers of Swedish are often utterly confused
when having to differentiate between tro, tycka and tänka, because they only have one word
for these concepts. When semantic spaces do not correspond between two languages, as in the
case of think/tro, the tendency is for L2 speakers to overextend the semantic boundaries of
their L2 to “match” their L1. A comparison of semantic boundaries between L1 and L2 items
shows that semantic boundaries which are similar are convergent, while semantic boundaries
23
that are dissimilar (think/tro) are divergent. The more similar two languages are, the more
convergent these semantic boundaries will be. Divergence is nearly always caused by moving
from a “coarser” to a finer “grid,” e.g. moving from the L1 item tro to think/believe
(Hasselgren 1993: 54). Such types of semantic lexical transfer are known as finer gridding –
when the transition from a coarser L1 (Norwegian) grid to a finer L2 (English) grid causes the
wrong word to be chosen.
However, evidence from this study shows that even non-semantic aspects of words are
transferred from the L1 to the L2. This is reflected by collocational transfer (collocating
words based on how they combine in the L1), stylistic transfer (words that are perceived to
have a 1:1 relationship between L1 and L2, and which are stylistically appropriate in the L1
and not the L2), syntactic transfer (tacking syntactic features of an L1 item on an L2 item)
and the creation of non-existent phrases or words in the L2 that derive from the L1 (one
student in my study wrote office rat (kontorrotte) to denote people who spend too much time
at the office). Because of these instances found in my data, I would suggest that the semantic
equivalence hypothesis be stretched to a word equivalence hypothesis – if an L2 speaker
perceives and L1 and L2 word as having a 1:1 mapping, they will often transfer knowledge of
their L1 to alter the semantic space of an L2 item, its word combinations, stylistic markings
and even syntactic features.
The second major type of lexical transfer is loan translation. Loan translation is “the
use of the literal translation of compound words, lexical collocations, or idioms from one
language to another” (Ellis 2008: 370). This will be further explained in section 5.2.1.2.
4.2.1.2 Weak L1-influence
Finer gridding can be the result of lexical transfer (such as in the case of a Swede overusing
think), but not necessarily so. For instance, how can semantic lexical transfer explain if a
student overuses believe instead of think? Which of the L2 items will the L2 speaker pick as
the “overextender?” In cases where L1 items are subdivided into two or more items, such as
tro, which has components common to think but also additional meaning (believe), it has been
observed that the learner will “intuitively apply the most familiar of the L2 items to the
whole semantic area in question” (Hasselgren 1993: 15, emphasis mine). A Swedish speaker
of English may therefore prefer think because it is more similar to tro in form and main sense,
or s/he may prefer believe if this is the L2 item s/he has heard the most and is the most
familiar with. In the case of think/tro, picking think indicates a strong L1 influence due to
24
semantic transfer, because the student (most likely) perceives a 1:1 mapping between a
familiar item (think) and an L1 item (tro). In the case of believe/tro, believe indicates a weak
L1-influence, because the problem lies in having to divide between semantic spaces that do
not exist in the L1, rather than transferring semantic aspects of the L1 to the L2. The
incorrectly picked word was not influenced by the L1. Instances of finer gridding where an L2
item that is not similar to the L1 is picked is therefore termed weak finer gridding, and
labeled as weak L1-influence.
Another instance of weak L1-influence is the overuse of basic words (known as “core
words” in this thesis; see section 5.2.1.6). Often basic English words will have a Norwegian
equivalent (e.g. big – stor). When a Norwegian speaker of English chooses to overuse big, it
is more likely preference for a basic word, rather than an L1-influenced choice of words.
However, basic words in English and Norwegian behave differently. For instance, big is seen
as much more informal, such as big question. It has an informal stylistic mark, so L1 English
speakers will prefer synonyms that have a more formal mark when writing formal texts, such
as large. Norwegian stor does not have an informal stylistic mark, so writing stort spørsmål is
more acceptable in academic writing. Using basic words in academic writing shows clear
dissonance in English, but not to the same degree in Norwegian. Therefore, (some) basic
words have weak L1-influence: it is possible that Norwegian speakers of English find them
more acceptable in English because they are acceptable to use in their L1. Instances where
basic words have an equivalent L1 that does not have the same stylistic markings are
considered to be weakly L1-influenced. As in the case of weak finer gridding, the problem lies
in having to divide the language differently from the L1. These instances are not considered
intralingual influence, because if the L1 item had had a finer grid, the particular L2 item may
not have posed a problem.
4.2.2 Intralingual influence
Intralingual influence is interference or confusion within the target language itself.
Intralingual influence is unrelated to the L1 and can either reflect unique errors of the
individual speaker or the developmental universal processes that are a part of language
acquisition in general (Ellis 2008: 53, Ortega 2009: 51).
In vocabulary acquisition/use, developmental processes are often categorized by level
of meaning, e.g. from concrete to abstract senses of words. A common intralingual problem is
overgeneralization. L1 children overgeneralize concepts, such as calling all animals with four
25
legs “doggy.” However, L2 speakers have most likely developed these concepts in their first
language, and overgeneralization of concepts tends to be replaced by L1-influenced semantic
lexical transfer. L2 speakers overgeneralize by overusing superordinate, approximate terms, in
instances where a native speaker would have used a more specific term , for instance to do a
study, instead of conduct a study (Hasselgren 1993: 8). Wrong L2 words that have no
perceivable L1-influence – especially between two languages that have little vocabulary in
common – tend to be viewed as a result of intralingual influence. What causes these
intralingual errors is unclear. López explains them as being a consequence of “partial
exposure to the target language” (López, date of publication unspecified). Hasselgren stresses
the challenge of choice that L2 speakers are faced with when “selecting words from an often
confusing and incomplete stock of L2 words in the interlanguage” and refers to this selection
as “lexical coping” (Hasselgren 1993: 1). Linguists who study vocabulary acquisition, such as
Hasselgren (1993, 1994), Schmitt (2010) and Martin (1984), all seem to be in consensus that
picking the wrong word is the result of a) lack of word knowledge and b) coping with lexical
choices when there is a lack of word knowledge. Sometimes these lexical choices will be
influenced by the L1, intralingual factors, a combination of the two, or other factors that
researchers cannot yet explain.
In this study, words that do not have perceivable L1-influence are considered a
consequence of intralingual factors, which could be the result of overgeneralization, the
picking of the wrong synonym (arbitrary guessing or confusion of word forms), teaching-
induced errors (students are encouraged to use overuse certain terms) and lastly, idiosyncratic
errors which are most likely the result of a conceptual misunderstanding of the word within
the individual speaker's mental lexicon.
4.2.3 Errors – interlingual or intralingual?
One of the heated topics in SLA literature and research is how large a role interlingual
influence – particularly L1-influence – plays in acquisition and use of the L2. The role of L1-
influence was stressed in earlier times, such as the rise of contrastive analysis in the 1950s-
1960s. Contrastive analysis stated that differences between the L1 and L2 could purely be
explained through a systematic comparison of the two languages. Any differences between the
two languages could predict possible problems for L2 learners. The contrastive analysis
hypothesis stated the following: in cases of similarity between native language and target
language, the learner could just transfer their native tongue habits. If this transfer was
successful, this would then be positive transfer. In cases where the two languages differ,
26
negative transfer (a structure that exists in your native language but not in the target
language) could be predicted. However, the L1 cannot always account for the mistakes L2
speakers make. Analyses began showing that a large proportion of errors could not be
predicted by contrastive analyses (Simensen 1998: 91).
With stacking evidence against the deterministic contrastive analysis, a weaker version
of contrastive analysis, where only some errors could be explained by comparing the two
languages, was suggested. This focus on explaining the source of errors was the beginning of
error analysis, and examined both interlingual and intralingual errors. In modern times,
intralingual influence has been studied and is more widely accepted as a large factor in L2
errors. Nevertheless, the dichotomy between interlingual versus intralingual influence causing
the most amount of confusion and errors in L2 is still widely discussed. For instance, Ellis
writes that “A large number – and in some cases perhaps most – of the errors that learners
produce are intralingual in origin rather than transfer,” but concludes that “the proportion of
transfer and interlingual errors varies in accordance with the task used to elicit samples of
learner language” (Ellis 2008: 55).
Yet the divide between inter- and intralingual influence is not as clear-cut as presented
above. Errors can have more than one source, as illustrated in Hasselgren's treatment of
influence on choice of wrong words. She finds the need to create an L1/L2 category which
covers cases where she sees that “both languages seem to exert a simultaneous influence”
(Hasselgren 1993: 46). She provides an example of an L1/L2 combination in the case of when
one student chooses overlook as an appropriate translation of overse (ignore). Here there is a
dual factor of literal translation and a word association error (see section 5.2.1).
Since errors can have more than one source, and it is impossible to know for certain
what the speaker is thinking when producing the errors, several linguists are skeptical of
labeling sources. Flick (1979), for instance, comments that assigning sources to errors is
“largely an arbitrary matter, subject to the individual biases and point of view of the
researcher” (Flick 1979: 60, quoted in Ellis: 55). Nonetheless, this thesis aims to point out
potential sources of errors.
5. TAXONOMIES OF LEXICAL ERROR
There are many taxonomies of lexical error in SLA literature. Ostensibly, these taxonomies
27
have little in common. Llach, in an attempt to characterize them, lists eight main classification
criteria, (Llach 2011: 76-87):
1) Distinction between form-oriented and content-oriented lexical errors to explain if
the lexical errors semantic or formal in nature.
2) Descriptive criterion includes pure descriptive considerations that focus on the
surface form of the error without alluding to causes or sources. Examples: wrong lexical
choice, omission, wrong order.
3) Etiologic or psychological criterion explains the source of lexical error. Lexical
errors are classified relative to their cause – mental processes underlying the lexical error.
Example: overgeneralization, semantic transfer and confusion of related words
4) Origin of influence criterion includes interlingual, intralingual and teaching-induced
errors .
5) Grammatical or linguistic criterion has to do with the classification of lexical errors
depending on the linguistic level on which the error occurs. Phonology, orthography,
morphology, syntax.
6) Word class criterion examines which word classes are the most affected by errors.
7) Product-/process-oriented taxonomies take psychological processes that generate
errors into account.
8) Miscellaneous. Taxonomies that combine several classification criteria in an
endeavor to establish a complete taxonomy that collects as many lexical error types as
possible.
Taxonomies are in most cases developed to fit the data – not vice versa. Therefore, it is
difficult for lexical error taxonomies to be adapted for other data sets, such as learners with
different L1s, or different types of data (collected through elicitation method, judgment tasks
etc) (Llach 2011: 91). Due to these issues, it seemed most prudent for this study to use a
taxonomy of lexical errors created by a researcher who had similar data. The first criterion
was that it should be a taxonomy created specifically for the Scandinavian L1s (Norwegian,
Swedish or Danish), since the languages are similar to each other (for comparison) and similar
to English (for taxonomies that focus on similar L1-L2s). A major figure in lexical error
research within the Nordic countries is Håkan Ringbom (1987, 1992, 2007), who investigates
28
L2 English lexical errors in Swedish- and Finnish-speaking Finns. However, Ringbom only
takes into account interlingual influences, and thus intralingual influences are not explained.
His taxonomy was therefore not ideal. Hasselgren (1993), on the other hand, created a
taxonomy for Norwegian speakers of English that also includes intralingual influence. Her
taxonomy is of Llach's “miscellanous” type because it features criteria such as content-
oriented, psychological (routes), origin of influence (L1/L2 influences) and descriptive
(effects). A few modifications were needed to fit the taxonomy to my data since her data
consist of translation texts and mine consist of free production texts. How the taxonomy is
altered will be addressed in 5.2. Hasselgren does not start from scratch – her taxonomy has
origins in Marilyn Martin's taxonomy of dissonance. Therefore, the next section will examine
Martin's and Hasselgren's taxonomies of lexical dissonance, accounting first for Martin's, and
how Hasselgren built on Martin's model.
5.1 Martin's taxonomy of lexical dissonance
Marilyn Martin wrote an article in 1984, titled “The problem of synonyms.” In it, she attempts
to explain why some advanced speakers sound “foreign” in relation to vocabulary. Martin
concludes that dissonance arises when a student has made an incorrect hypothesis about the
wrongly chosen word, e.g. that an informal word can be used in formal contexts (Martin
1984: 130). According to Martin, dissonance occurs first and foremost because students are
not provided with enough information of new words in the foreign language classroom, be it
glossaries or “synonyms.” For instance, a Norwegian student of English who is taught that
large means “the same” as big may be in for a surprise when they want to tell someone about
their large sister (storesøster). The wrongly chosen words are synonyms that share some
properties with the correct word (large – big), but there are several aspects of words that make
synonyms different from one another. When L2 speakers are led to believe that such
synonyms are completely interchangeable, dissonance takes place. Martin proposes that there
are four types of dissonance that interfere with encoding by advanced learners, and once these
dissonances are identified, language teachers will more easily be able to pinpoint why a word
is “wrong” and furthermore teach words within this framework to avoid further mistakes. The
four types of dissonance are4:
4 These types of dissonances are known as “effects” in this study, and will be further explained in 5.2.2. Martin
presents a highly detailed list explaining her taxonomy. As these categories are treated slightly differently in
Hasselgren's taxonomy, these will not be explained in great detail, but can be found in Martin's article, pages
131-134
29
I) Semantic (the speaker has misunderstood the meaning of the word, e.g. superordinate/
hyponym)
II) Stylistic (the speaker has misunderstood the style of a word, e.g. formal /informal/ taboo)
III) Collocational (the speaker is unaware of combinations of prefabs)
IV) Syntactic (the speaker is unaware of the words that tend to surround the dissonant word,
e.g. case relation confusion)
Martin suggests that if these four types of dissonances are explained in detail to the language
learner, many synonymous errors can be avoided, and ideally L2 speakers can teach
themselves these different aspects of word knowledge, to avoid habitually using words
interchangeably.
Martin's taxonomy for lexical dissonance is a good starting point for any lexical
researcher because it is a comprehensive taxonomy that can account for why several types of
words can be dissonant. However, there are several shortcomings with her model, i.e. there
are dissonances between meaning and context that fall outside the framework of her
categories, and thus cannot be explained by them. Carter points out among other things that
grammatical and lexical components combined will also result in dissonance, e.g. He ran very
fast and failed to win the race (Carter 2012: 83). Additionally, dissonance does not pertain
only to the clause-level. Carter illustrates this with beyond-the-sentence errors, such as: She
passed the exam. This move pleased her parents (Carter 2012: 83). He explains this as:
To account for and understand lexical patterning at the level of the clause, ways with
words depend just as crucially on the patterns created by lexical items in the wider
context of naturally occurring discourses (Carter 2012: 83).
Another problem with the taxonomy is the assumption that only wrongly chosen synonyms
can account for lexical dissonance. Occasionally an L2 speaker will use a word without
knowing it at all, or use a word and perceive it to mean something completely different. The
end result is at best a word that is unrelated to the right word yet guessable through context,
and at worst an entire clause/sentence where it is impossible to understand what the speaker
means. An example of this is my definition of “perceived equivalents” (see 5.2.1.3), where L2
speakers will use words that cannot be explained by any of the suggested incorrect hypotheses
in Martin's framework. For instance, one student writes: A division in time. This could
possibly mean gap if we look at the surrounding words and our knowledge of words that
30
collocate with time – but division and gap are not synonyms. If gap is not what the student
meant to express, I have no further suggestions. Wrong words in this category are found even
at the advanced level of Norwegian speakers. The reason for why unrelated words can be
picked instead of synonyms is that failure to encode the word (e.g. a form-linkage failure) can
cause students to arbitrarily pick a word they do not know at all, instead of simply using
“similar” words interchangeably.
Martin's neglect of L1-influence also results in a lack of coverage of dissonant items.
L2 speakers are almost referred to in her article as less knowledgable L1 speakers, rather than
second language learners who have an abundance of knowledge and resources to call upon
when they are in doubt. It is impossible to claim that L2 speakers only refer to the second
language when speaking/writing in the second language – they often draw on knowledge of
their native tongue when they are unsure of a new word, and create “incorrect hypotheses”
based on their L1 knowledge, e.g. semantic lexical transfer. Seasoned second language
teachers such as Hasselgren recognize the importance of the L1 – even in advanced speakers
of English. The semantic equivalence hypothesis is an example of this – when some aspects
of words are unknown, L2 speakers rely on the structure of their own language (high - høy),
rather than transferring knowledge of other, similar L2 synonyms (high – tall).
5.2 A taxonomy of lexical dissonance for Norwegian students of English
There are several weaknesses with Martin's taxonomy that need to be addressed if it is to be
practically implemented. First, it is purely descriptive in nature because it does not explain the
sources or mental processes behind choosing the wrong word. Another shortcoming is that
L1-influenced dissonances which may create items that do not exist in the L2 (e.g. office rat)
cannot be explained by the taxonomy. In Hasselgren's examination of Norwegian student
texts, these shortcomings needed to be addressed in order to answer the following questions:
i. Why do learners choose wrong words? (Influence)
ii. How do they choose wrong words? (Route)
iii. What is wrong with the choice of words? (Effect)
31
Such an investigation needs among other things “origin of influence” and “psychological”
criteria to explain the how and why in the choice of wrong words. Therefore, Hasselgren
creates her own taxonomy for “routes” (how learners choose wrong words, or psychological
criterion) and “divergence” (why do learners choose wrong words, or origin of influence), in
addition to Martin's taxonomy, which she labels “effects” (what is wrong with the choice of
words). The process from choosing a wrong word to its effect can be explained in the
following way by Hasselgren's taxonomy: a learner does not know how to express tall, as in
he is tall (han er høy). The learner knows Norwegian høy, which has an English cognate,
high. The learner also knows that high is an English word, and assumes a semantic
equivalence, i.e. that høy can be directly translated to high. The learner then chooses this
cognate as a “route,” or pathway, and the learner chooses the wrong word because s/he relies
on her/his knowledge of L1 (Norwegian). The belief that høy = high is an incorrect
hypothesis, because high is not used to describe people in English. The effect of the wrong
word is semantic dissonance.
According to Hasselgren's taxonomy, Norwegian speakers of English use six routes
(pathways that can potentially lead to dissonance), and there are five different ways in which
a word is wrong (effects). The dissonances can be caused by L1-influence, L2-influence,
L1/L2 influence or other influences (e.g. obvious avoidance of certain terms). The next
sections will outline routes, effects and influences in detail. Since Hasselgren uses translation
texts and I use free production texts, I found the need to modify and add to some of her
categories.
5.2.1 Routes
What is a route?
A route is how learners make mistakes, or in what way a learner selects a wrong word
(Hasselgren 1993: 45). Hasselgren created the taxonomy “routes” herself by examining large
bodies of texts written by Norwegian speakers of English and determining the different ways
students select wrong words. She posits that there are six ways in which Norwegian students
of English misselect words: cognates, transliterations, perceived equivalents, synonyms,
associations and cores. Some of these concepts are by no means new: Ringbom has been
investigating the role of cognates and transliterations in learner language since the early 80s.
However, Hasselgren is the first to build a system that could explain all possible ways
Norwegian speakers of English misselect words.
32
5.2.1.1 Cognates
A cognate is "a linguistic form which is historically derived from the same source as another
language/form" (Crystal 2008: 83). An example of this is English see, which has a Norwegian
(historically derived) equivalent: se. In many cases, cognates lead to positive transfer – they
are easier to acquire and use, such as in Ringbom's studies of Finnish versus Swedish speakers
of English (1987, 1992, 2007), where the Swedes have a great advantage over the Finnish. He
explains this as: “when both phonological and semantic similarity work together, the effect is
like that of a magnet attracting a new word to be stored in the learner's mental lexicon when
he meets it for the first time” (Ringbom 1987: 41). In other cases, they are misleading, such as
when they cover different semantic spaces or collocate differently. Equivalent cognates lead
to positive transfer and facilitate second language learning, whereas non-equivalent cognates
lead to negative transfer. This is stressed by Ringbom, who discovered that false friends (a
Swedish and English word that have formal similarities) were the highest source of errors due
to Swedish influence in his study (Ringbom 1987: 124).
Cognates always have L1-influence, which means that they are L1-based. A wrong
word in this study is labeled a cognate if the L2 item has an equivalent L1 item that is similar
in form and (occasionally) meaning. Examples of this are: train – trene (I train at the gym),
mean – mene (I mean that fruit is healthy), clock -klokke (What is the clock?) and land - land
(Norway is a nice land). If the word is historically derived from the same source or not is not
a criterion (although in many cases it will be). My reasoning is that when a Norwegian
speaker of English decides to perceive two words as being the same in both languages, they
do not consider whether the words are linked together historically, but rather if the forms and
meanings are close enough for there to be a 1:1 relationship between the two. Words that are
similar in meaning but not in form are considered transliterations (for instance, live - bo).
5.2.1.2 Transliterations
Transliterations, much like cognates, stem from L1-influence. Hasselgren explains
transliteration as: “the process of breaking down an L1 item, and reassembling it with L2
parts, often resulting in a non-existent word or phrase” (Hasselgren 1993: 48). Sometimes
breaking down an L1 item will result in positive transfer – particularly if the two languages in
question are similar. Any instance where you correctly guess a word or expression in a foreign
language with L1 influence is an illustration of successful transliteration. In cases where L2
knowledge is lacking and a guess based on L1 knowledge is made unsuccessfully, dissonance
33
caused by transliteration occurs. The main difference between cognates and transliteration is
that if the form of the two words is similar, they are classified as cognates in this study, and if
they are directly translatable words that are not similar in form (fin - nice), they are classified
as transliterations. Examples of transliteration are: to live at the hotel for the weekend (bo på
hotell), to make a party (lage en fest), and to do a good try (gjøre et godt forsøk).
5.2.1.3 Perceived Equivalents
Perceived equivalents are defined by Hasselgren as:
a small but widely used group of L2 words felt instinctively to be chosen under the influence of particular L1 items, yet having no formal resemblance to them. An example is the verb experience, frequently provided when find out or discover would be more appropriate (Hasselgren 1994: 241).
Perceived equivalents in this analysis are treated differently. The reason for this is that
Hasselgren had access to Norwegian source texts. In free production texts, there is no such
insight into what the student was supposed to or wanted to express. It is therefore impossible
to determine what the student believes to be a 1:1 mapping, unless the student repeats the
same mistake several times in different contexts, showing that when they write experience,
they clearly mean discover. This analysis includes such repeated errors where context shows a
false 1:1 mapping between the Norwegian L1 item and the English written item. Such
repeated errors do not have any perceivable L1-influence in my study, and are often explained
by unsuccessful coding of the L2 word, or other unknown factors. The criterion for a
perceived equivalent in this study is a wrong word that is repeated several times, and where
there is no perceivable L1-influence. An example of this is when one student uses the word
profit instead of goal/aim twice in a text. Example: Maybe the profit is to reach popularity.
Profit and goal/aim are not related in any way, either semantically or by word form. There is
no perceivable influence from the L1 because Norwegian differentiates between profit
(profitt/overskudd) and goal/aim (mål), and neither of these L1 items should be confusing to
the L2. The student also repeated this error 2 times, indicating that s/he perceived profit to
mean goal.
Perceived equivalents also includes words where the intended meaning is simply
unknown. Examples of unknown intended meaning are: she gave him a withering look, or a
mortal glance. The reason why these unknown dissonances are considered to be perceived
equivalents is because there is an underlying assumption that when an L2 speaker produces a
sentence, they have a clear picture of what they intend to say (in other words, they do not just
34
write words arbitrarily without some intended meaning they wish to convey), whether they
have a Norwegian L1 item in mind or not. So while the student perceives the word they write
to have the meaning they wish to express, either the word form or sense is wrong in some
way, and the intended meaning cannot be determined (unless it becomes clear through
context).
Since this route does not have perceivable L1-influence in my data, it is labeled as L2-
based. The criterion for an L2-based route is that the difficulty lies in coping with the choices
and restrictions of the L2, rather than interference from the L1 (Hasselgren 1993: 50).
However, as in the case of perceived equivalents, problems could also mainly lie in the
encoding of information of the word.
5.2.1.4 Synonyms
Synonyms are words with approximately the same meaning. However, no two words can
mean exactly the same. The fact that words can be seemingly similar yet differ on many levels
is a point stressed by Martin, who created the whole category “effect” to explain why
choosing a synonym can mean choosing the wrong word.
Categorizing the route “synonyms” is tricky , for are not most dissonances the result of
simply picking the wrong synonym? “Synonyms” is therefore a miscellaneous category for
dissonant words where the intended meaning is clear, but there is a) no direct L1-influence
(cognates and transliterations), b) the wrong word and correct word are not similar in form5
(e.g. take an assumption instead of make an assumption) and c) the synonym is not an
overused basic word where several other correct items would have been appropriate6 (e.g.
nice instead of wonderful, splendid, fantastic). Synonyms belong to the same word class and
often the same semantic or associational field (i.e. the words are related in meaning or
grouped together by how we associate them, e.g. pillow – bed).
“Synonyms” is L2-based because the problem does not lie in direct transfer of word
knowledge from the L1 to the L2. Synonyms can also have L1-influence, but this influence is
weak, as in the case of weak finer gridding (see section 4.2.1.2). Unlike the L1-based
cognates and transliterations, many synonymous mistakes bear no resemblance to Norwegian
translations/equivalents. Examples of synonyms: the volcano detonated (erupted), or a smile
5 These dissonances are labeled “associational” dissonances
6 These dissonances are labeled “cores”
35
crossed her mouth (lips).
5.2.1.5 Associations
Associational dissonance is dissonance where a) the L2 speaker has chosen an incorrect
synonym that is similar in form (and possibly meaning) to the correct word or b) the student
has spelled a word so incorrectly it somehow changes the form of the word (such as sward*
instead of swore). Association bases itself on the dual factors of form and (loose) meaning
association, i.e. words that sound alike or are written similarly are confused with one another
when selecting a synonym (Hasselgren 1994: 242). This confusion of form and (possible)
meaning association does not only pertain to L2 speakers. Learner errors caused by
association with phonologically related words (take - make) are strikingly similar to native
speaker “slips” (Meara 2009: 28). Selection errors (a type of “tongue slip” where the wrong
item is selected from the mental lexicon) can be based on meaning similarity (He came
tomorrow (yesterday)), sound meaning (The emperor had several porcupines (concubines))
or meaning and sound (I don't have empathy for rich-looking burglars (beggars)) (Aitchison
2012: 22). In this study, associational dissonance is based on sound similarities or a
combination of meaning and sound. Sound similarities are called malapropisms – cases in
which a similar-sounding word has been wrongly selected, as in cylinders for syllables
(Aitchison 2012: 158-159). Words that have a meaning similarity (tomorrow - yesterday) are
placed in the synonym category.
Associational dissonance also takes the relationship of lexemes (pertaining to word
families) into account, such as to hold a speak (speech), or her hair looked beautifully
(beautiful). If the wrong word class for the correct lexical item is chosen (speak - speech),
these are labeled associational dissonance, since it could be the word form that is confusing.
It is difficult to determine where spelling mistakes end and associational dissonance
begins. Accidental or common spelling mistakes (such as “slowe” for “slow”) should be
disregarded in lexical studies. Only dissonance that reflects a lack of knowledge of word
sense, collocation, syntax or severe misspelling of word forms are considered. Several
spelling mistakes which can be found in the texts are therefore omitted from the list of
dissonances. Examples of associational dissonance include an exaggerated birthday party
(extravagant), to increase your looks (enhance) and taught instead of thought. Associational
dissonance is an L2-based route, because the wrong L2 item is not influenced directly by an
L1 item.
36
5.2.1.6 Cores and Lexical Teddy Bears
Cores and lexical teddy bears are words that block the production of more appropriate words
or a variety of synonyms. Cores and lexical teddy bears are not exactly the same, but both
have the same effect of blocking production and often overlap. This section will explain what
core words and lexical teddy bears are, why they are dissonant and how they are classified.
5.2.1.6.1 What is a core word?
As per definition, a core word is a neutral word with many senses that can collocate freely
and is neither too formal or informal. Carter furthermore defines core words as: “central
members of their lexical sets or superordinates, and they will collocate widely, frequently
being extended into phrases” (as cited in Hasselgren 1993: 33). Core words often stem from
core vocabulary – a group of widely used words that are more resistant to loss or change than
other parts of vocabulary (Saeed 2009: 77). Normally, core vocabulary is established and
based on frequency studies, particularly with the aid of corpora in recent times. Mapping core
vocabularies is thought useful for two main reasons: 1) establishing how similar or different
languages are and 2) to boost speed in L2 acquisition. Core words are considered ideal for
quickly improving second language vocabulary, as they have many notions which can be re-
expressed, and are often considered “basic” language (Carter 2012: 40). One project which
attempted to map core words for such a purpose was the Basic English project, which sough
to provide a basic minimum vocabulary involving a list of 850 words (Carter 2012: 37).
However, projects such as the Basic English project often fall short, for even though it seems
simple enough to learn 850 words in order to speak basic English, Nation points out that these
850 words can have as many as 12, 425 meanings (Nation 1983: 11, quoted in Carter 2012:
40), and research has not established whether it is easier to learn the extended senses of such
core words, or to learn the different meanings of an almost equal number of lexical items
(Carter 2012: 41).
5.2.1.6.2 Core words – wrong?
Even though core words are often characteristic of L2 learners and considered dissonant in
some cases, they are by no means “wrong” to use. Native speakers make frequent use of core
words, which is why they are considered “core” in the first place. A core word is only
considered dissonant when a student uses a core word when a more specific, register-
appropriate word would have suited better. For instance, the adverb pretty and adjective good
37
are perfectly appropriate in colloquial English, such as pretty good results. However, in an
academic setting they are not register-appropriate, and should be replaced with e.g. quite
positive results, or other synonyms that reflect the style of register. The reason why core
words are not labeled as synonymous dissonance (which, of course, they are to some extent)
is because core words reflect an interference from words learned early in the L2 stage. For
instance, an L2 speaker who learns good at an early stage will be more reluctant to use or
learn other, more specific synonyms, such as admirable, satisfactory, or splendid if their use
of good feels adequate and is positively reinforced. Hasselgren explains this as:
words with highly specific meaning and restrictions on use are unpopular - got will usually be preferred to acquired […] Multiplicity of meaning has been shown to lead learners to avoid using a word in more than one sense, a point reinforced by Laufer [...], who cites polysemy as the largest single source of comprehension errors (Hasselgren 1993: 19).
Overuse of core words is often characteristic of non-native speakers, as they lack some of the
lexical nuance and density that a native speaker is more likely to possess. This is evident in
Hasselgren's own study of “different words,” where she concludes that:
learners have a distinct preference for “core items.” While native speakers tended to choose intensifiers that somehow capture the specific semantic area, Norwegians on the whole did not do this. When native speakers favor unrestricted items, such as great or pure, the learners tended to follow suit. However, when the favored native item was restricted, such as profusely or utterly, the two groups parted company (Hasselgren 1994: 254).
5.2.1.6.3 Labeling core words
Core words, particularly stylistic ones, are difficult to label. The reason is not because it is
difficult to determine whether a core word is a core word or not – the difficulty lies in
determining whether these core words are inappropriate or not in context, and if another
lexical item would have been used by a native speaker. Some of the intermediate free
production texts are literary. Core words are largely not considered dissonant in this genre,
because the register is appropriately informal and colloquial. Even in academic writing there
is a gray area between when a core word is considered inappropriate or not in context. For
instance, very is generally considered inappropriate and is avoided in essays, such as in the
example: a very interesting article. Most native speakers are aware of the redundancy of using
this adverb, and will either avoid it, or use a stylistically appropriate synonym, such as
fascinating. However, this does not mean that very is always dissonant in argumentative
essays. In another instance, a student wrote: these are shown in very different ways. In this
case, very is not redundant and adds to the text. When the core word list states that there are
38
38 instances of very as a core dissonance for intermediate students, it does not mean that the
word very is only used 38 times – each context has to be taken into account to establish if the
core word shows traces of dissonance or is used appropriately. So if a core word struck me as
being inappropriate in context, it was labeled as dissonant.
5.2.1.6.4 Lexical teddy bears
The term “lexical teddy bear” has been used without much explanation. First coined by
Angela Hasselgren, it is explained as: “stripped of the confidence and ease we take for
granted in our first language flow, we regularly clutch for the words we feel safe with: our
'lexical teddy bears' (Hasselgren 1994: 237, emphasis mine). This sense of “clutching” to
words is considered a type of lexical dissonance because a word will become much more
frequent than in native speaker use. The words we cling to in our second or third language are
more often than not core words learned at an early stage, but not necessarily so. Occasionally
we simply find words that we like, feel comfortable with and overuse. An example could be
overuse of the word really (really good, really great), which does not have the neutral status
of a core word (it is stylistically marked as colloquial), yet is overused. The danger of lexical
teddy bears is that they may lead to imprecision (he never managed to take the exam), wrong
collocations (this didn't awake any interest) which results in production blocking of other,
more accurate words (Hasselgård et al 2007: 66). An example of a core word used as a lexical
teddy bear: The histories [stories] tell us a very good story […] I would say that if you live
with a thread it's hard to live a good life […] I think that is good that some people write about
this type of stories.7 Here, a multitude of synonyms could have been used, but the student
clutches to the core word good to describe a wide range of senses. Although core words can
be and often are overused, this is not necessarily the case. For instance, the core word bad is
only found twice in the intermediate student texts. Since bad is stylistically incorrect, it is
considered a core dissonance. However, since it is used so infrequently, it is not a lexical
teddy bear. Setting a definite limit for when a word is a lexical teddy bear is a difficult task –
it would be easier to put a word on a continuum where it is either frequently or infrequently
used compared to L1 use. However, a limit had to be set for the purpose of this study. A
lexical teddy bear is therefore a stylistically inappropriate word used over five times in a
corpus (25 000 words), with or without being a core word.
7 Example taken from one of the students in the intermediate corpus
39
5.2.2 Effects
What are effects?
Once it is established how learners make mistakes, another question arises: why is it wrong?
Another way of asking this is: what effect does the wrong word have on the message? Is a
message downright wrong in meaning, or simply wrong in style? For each route (cognates,
transliterations, perceived equivalents, synonyms, associations and cores), there is a potential
for each one to be dissonant in five different ways (although not every route and effect can be
combined, and some route/effect combinations are more common than others).
Martin (1984) was the first to propose how (similar) words can be wrong in context,
through four different types of lexical dissonance: semantic, stylistic, collocational and
syntactic. To this, Hasselgren adds an additional category to cater to L2 speakers: invalid.
Martin's account of effects is highly technical and does not explain how these effects work in
an authentic context. Hasselgren's definitions of effects offer a more surmountable way of
labeling dissonances.
5.2.2.1 Semantic
Hasselgren writes that: “semantic dissonance occurs when the meaning of the L2 item is
completely different from the L1 source item” (Hasselgren 1993: 56, emphasis mine). This
could be for instance semantic lexical transfer: when a student applies the senses of an L1
word to an L2 word. Since free production texts do not have an L1 source item, much like in
the case of perceived equivalents, this definition is not helpful for my study. The only way to
pinpoint semantic dissonance in free production texts is through context. So in my case,
semantic dissonance is judged based on contexts where there is a mismatch between the word
used, the possible senses that word could have and what I perceive to be the student's intended
meaning based on context. In other words, in this study, semantic dissonance is a mismatch
between L2 senses and L2 context, where the “right” word can or cannot be guessed through
context. Examples of semantic dissonance found in my material include: Her arms were
latent on each side of her (lying), or it is not immoral to do some changes or improvements
(bad/negative).
40
5.2.2.2 Collocational
Collocational dissonance is defined as a word that is “inappropriate simply through
disharmony with the words around it without being deviant in meaning or style” (Hasselgren
1994: 243). Unlike some semantic dissonances, collocational dissonances and their
appropriate equivalent are discernible in free production texts because the dissonance is
between the node word and its collocate (see section 3.2.3). Examples include: An evil circle
(vicious) or to squeeze a needle (inject).
5.2.2.3 Stylistic/Connotational
Stylistic/connotational dissonance covers items that are semantically correct, but have an
inappropriateness of style or connotation (Hasselgren 1994: 243). The reason why the
categories stylistic and connotational are lumped together as one effect is that they frequently
occur together. Here there is a mismatch between Hasselgren and Martins' taxonomies.
Connotational words are labeled as semantic in nature by Martin. In this case I agree with
Hasselgren – although there is a semantic aspect to connotation, the phrase She persisted
helping him (continue) will strike a native speaker as having strange connotations more than
being an actual misunderstanding of the sense of persist.
This study differentiates between informal, formal and connotational dissonance.
Informal example: Working with sick people may seem gross (unpleasant). Formal
example: Important for a human to have a job (person). Connotational example: She was
too busy [eating] the cadaver (dead body). Words are considered connotational if the
dissonance is perceived as being more connotational in nature than belonging to the
formal/informal dichotomy.
5.2.2.4 Syntactic
Syntactic dissonance is dissonance where either the lexeme is right, but in the wrong word
class, for instance: they had an argue (argument). Hasselgren chooses to omit syntactic cases
where she considers dissonance to be grammatical rather than lexical in nature, such as
adverbs lacking the -ly ending. I also attempt to weed out syntactic cases that are a reflection
of grammatical rather than lexical features.
41
5.2.2.5 Invalid
This effect was created by Hasselgren to cater to items that do not exist in English. In her
case, this means words that are created by students through transliteration. Hasselgren offers
the example: cold degrees from Norwegian kuldegrader (Hasselgren 1994: 243). However, in
my analysis, I also found non-words that were not the result of transliteration. Synonyms and
associations also have non-words. Invalid seemed the most appropriate category for these
non-words such as new-mown grass instead of freshly cut, or the predictment of marrying is
that you want to be together forever instead of prediction.
5.2.3 Influence
A third question of interest in Hasselgren's study of lexical dissonance is: why do L2 speakers
make mistakes? Or, what triggers a lexical dissonance (Hasselgren 1993: 63)? Here the term
“influence” comes to play – “anything that is seen to cause a learner to make a wrong word
choice” (Hasselgren 1993: 44). As mentioned in section 4.2, linguists commonly differentiate
between two competing (and occasionally converging) influences: interlingual and
intralingual influence. Hasselgren uses “divergence,” which labels influence in terms of L1,
L2, L1/L2 and “other” (for instance, cases of obvious avoidance). Since our data are so
different, I found a need to create my own taxonomy for influences. This study divides
influence in three ways: influence found in the L1-based routes (strong L1-influence, as
described in 4.2.1.1), influence found in the L2-based (weak L1-influence as described in
4.2.1.2) and intralingual influence (see section 4.2.2).
Strong L1-influence is termed “strong” because it suggests that the dissonance is
caused directly by a Norwegian L1 item. In the L1-based routes, there is always a possible L1
item that is the root of the dissonance. For instance, if a Norwegian student writes: he is a
very high man instead of tall, it is reasonably certain that the dissonance is caused by the
cognate høy. Strong L1-influence is the result of the word equivalence hypothesis: the L2
speaker has assumed an equivalence between an L1 and L2 item on some level, and the
assumption that these aspects can be transferred from the L1 to the L2 has caused dissonance.
Another cause is loan translation, or L2 items which are combined in ways only possible in
the L1 (see 4.2.1.1 for more information)
Weak L1-influence is termed “weak” because it suggests an indirect possibility of the
dissonance being caused by a Norwegian L1 item. Here it is not the influence of a particular
word, but rather that the semantic space and general behavior of words is different between
42
the L1 and L2. These are normally instances of weak finer gridding (see section 4.2.1.2). For
Norwegian speakers of English, the difference in stylistic Norwegian/English core word
behavior is also considered weak L1-influence.
Intralingual (L2) influence covers instances where the dissonance cannot be
explained by L1-influence, i.e. all dissonances that have no perceivable L1-influences are
assumed to be L2-influenced. Examples include: A smile crossed her mouth (lips) or to quiz a
survey (take). Neither of these instances can be explained by Norwegian, so we must conclude
that there are other interferences at play, either within the L2 itself, or individual, personal
mistakes that have caused the L2 speaker to encode or perceive the word wrongly.
5.2.4 Shortcomings with the taxonomy
The taxonomy poses several problems, particularly for free production texts. The first
problem is determining L1-influence as a source of dissonance – there is less certainty in
predicting L1 interference, since no L1 item has been provided. The second problem is
overlapping found in some of the categories. Cognates/transliterations and synonyms
/associations are labeled differently due to word form – for instance, a dissonance is a cognate
if the L1 and L2 item are similar in form and meaning (history - historie) or a transliteration if
the L1 and L2 item are similar in meaning but not form (case - sak). However, how does one
judge when two words cease to be similar in form? For instance, break - bryte are cognates,
but not very similar in form. So is break a cognate or a transliteration? In these cases it was
personal judgment that decided which route was the most dominant/influential.
Another case in point is perceived equivalents. A dissonance is a perceived equivalent
if the correct L1 item cannot be guessed in context (e.g. a mortal glance). However, a more
skilled researcher may be able to understand what the L2 speaker meant in context, and thus
label it as a synonymous dissonance. Lastly, some of the dissonances can have up to several
routes and effects. For example, one student wrote: white-skinned people. This dissonance
could be associational (white - light) if we view it as a mistake of word form. However, in
Norwegian it is possible to say hvithudet (white-skinned), so this could be a transliterational
dissonance. If the dissonance is considered transliterational, the correct L2 item is white
people. Judging which dissonance belongs to which category in these ambiguous cases is up
to the individual researcher. The more personal a judgment is, the less scientific it becomes,
and studies conducted using the same taxonomy will have different results due to personal
bias.
43
Additionally, Hasselgren's taxonomy can only reflect instances of coping with lexis
when a dissonance arises. Routes and influence can also lead to successful lexical choices,
such as if the L2 speaker chooses high for the collocation high price (høy pris). We do not
have access to the student's mind, so there is no way of determining if the student has
“guessed” the correct lexical item. One way to counter this is to count instances of positive
lexical transfer (performance analysis) for a successful/wrong lexical choice ratio. Yet the
same problem arises, since it is impossible to determine if the student has “guessed” their way
to correct word choices as well. This is generally a problem when trying to etiologically
explain errors.
44
45
6. METHODS AND MATERIAL
This thesis uses error analysis to examine distributional patterns of lexical dissonance. The
taxonomy for classifying lexical dissonance outlined in the previous chapter will determine
the distributional patterns based on how often dissonance occurs in these categories. The
material in this study is taken from two separate corpora – one written by Norwegian students
at an intermediate level of English, and one written by Norwegian students at an advanced
level of English. The corpora are cross-sectional data, since they represent two groups in a
specific period of time for the purpose of comparison. This section will discuss the methods
and data used for this study, and additionally contrast some of the strengths and weaknesses
that follow with the methods and data chosen.
6.1 Discussion of types of methods and data
6.1.1 Error Analysis
Error analysis is a tool used for investigating how learners acquire their L2 (Ellis 2008: 45). It
is a method that has its origins in the field of SLA, and generally distinguishes between two
main tasks: investigating gaps in knowledge, and investigating how speakers cope with these
gaps (James 1998: 62). “Coping” in terms of lexis refers to communication strategies (ways of
communicating when a learner lacks a word) and learning strategies (ways of remembering
new words). Hasselgren believes that a third way of coping with errors is not by any strategic
move, but is “more a matter of stumbling” when trying to pick the right word (Hasselgren
1993: 43). This is what we refer to as routes. Routes can lead to lexical dissonance. By
identifying instances of “routes” that lead to lexical dissonance and examining how these
routes are distributed in overall lexical errors, this study hopes to find indicators of how
Norwegians cope with their English vocabulary.
Error analysis is undoubtably a useful tool, but has some limitations as well. Many
have pointed out (e.g. Simensen (1998) and Johansson (2008)) that error analysis only
partially reflects interlanguage. A full scope of what the learner both can and cannot do is a
better representation of interlanguage. Such a method is called performance analysis, and
studies the learner's overall performance, which would in this case include accurate use of
lexis and investigation of pragmatic failures (Simensen 1998: 94). Other limitations of error
46
analysis include the bias of evaluation judgments (researchers will judge errors differently),
lack of consensus of what an error is, and determining the source of an error (how can we
know for sure if an error is caused by interlingual or intralingual influences?) (López, date of
publication unknown).
6.1.2 Translation versus free production – pros and cons
There are two main types of data in error analysis: translation and free production. L2
translation data are texts which have been provided to the L2 speaker in their native tongue
(in this case, Norwegian). The L2 speaker translates the text to their L2 (in this case, English),
and the end result is a translation text. Translation texts are useful in investigation of learner
language because the translation data reveals a) how the learner expresses L1 words in their
L2, and consequently b) how the learner perceives the relationship between words in the L1
and L2. An advantage of translation data is that the L2 speaker's intended meaning (what
they are supposed to express) is available to the researcher. If intended meaning is available, it
is much easier to determine the L2 speaker's relationship between L1 and L2 words. For
example, if an L2 speaker writes discover in a translation text, a quick peek at the original
word oppdage will confirm that the L2 speaker has created an accurate 1:1 mapping between
the L1 and L2 word. Such insights into intended meaning are of particular aid to error
analysis, because sources of error can more easily be identified by seeing how the L2 speaker
perceives the relationship between L1 and L2 words, i.e., what the L2 speaker perceives a
certain L1 item to mean in their L2. Translation texts can therefore be said to more accurately
identify L1-influenced errors. However, translation texts are not without fault. Firstly, being
provided with L1 words is problematic on some level. The L2 speaker may have never used
the word provided in either their L1 and L2, and being provided “cue” words means the L2
speaker is limited in selection of words. Furthermore, translation texts focus on the L1-L2
relationship (how an L2 word is interpreted in relation to an L1 word), and not how an L2
word is perceived in an L2-context.
In Hasselgren's 8 translation tasks, 15 examples of each were provided, but they are all
a reflection of how 15 different students interpreted the number of words limited in eight
tasks. This can be considered an advantage in that for those limited number of words, there
are 15 different sets of data illustrating how learners cope with those particular words.
However, it is also a disadvantage because being provided cue words does not reflect how an
L2 speaker would naturally write about a theme, and the dissonances that follow free
production without L1-items.
47
In the 1990s, L1 (Norwegian) was seen as a positive factor in the English language
classroom (Simensen 1998: 123). Today, this is not the case. Translation tasks are therefore
perhaps a discipline belonging to the past, when relationships between the L1 and L2 were
more encouraged in the EFL classroom. With the rise of communicative competence and
immersion techniques, it is becoming frequently more popular in Norway to rely less on L1-
L2 glossaries and translation tasks, and instead allow students to use their own knowledge of
English in the EFL classroom and build on their L2 vocabulary based on the words they
already have acquired. Because of this, free production texts are more prudent to examine in
Norway – Norwegian students are no longer trained in comparing their L1-L2 knowledge, and
a comparison of L1-L2 knowledge is at any rate not the type of knowledge that is of interest
to current EFL studies in Norway. However, free production texts also pose a number of
problems in error analysis. First, there is no insight into the student's intended meaning, and
the only way to deduce errors is through context. Errors are therefore more prone to
interpretation based on the researcher. There is also a higher risk of avoidance of words the L2
speaker does not know, because there is no pressure to extract knowledge on specific L2
items. Lastly, there is less data on individual errors, because the participants can choose the
words they use themselves, so it is much less likely that participants will have the same types
of errors.
While Hasselgren (1993) uses translation data, this study draws on free production
texts. To my knowledge, Hasselgren's study is the only study that attempts to find
distributional patterns of lexical dissonances in Norwegian speakers of English. Initially, this
study sought to replicate Hasselgren's, to compare and support or challenge her results.
However, replicating Hasselgren's study was impractical due to the a) availability of
translations texts, particularly for intermediate students, and b) how relevant and natural
translation texts are for both intermediate and advanced Norwegian speakers of English, with
regard to the focus of the Norwegian educational system.
6.1.4 Cross-sectional data
The two corpora used for this study are cross-sectional data, meaning a set of data at a
specific point of time from a group that represents a whole. This is referred to as “snap-shot”
corpora, because they are set at a certain time and do not show participants' progression over a
longer period of time. The advantages of using such material is easy accessibility and
comparisons of several subjects at a specific point in time. A disadvantage is that cross-
sectional data cannot explain what will happen to the participants over time. Ellis calls this a
48
“static view of L2 acquisition” (Ellis 2008: 61). This study chooses to use cross-sectional data
since longitudinal data for Norwegian speakers of English from intermediate to advanced is
not available, and also because it wishes to draw comparisons between the two groups. Since
this data is not longitudinal, it cannot be determined if there has been any “progression” from
intermediate to advanced at an individual level. The participants are different, and the texts
are written at different times, allowing for differences in data due to variables such as chance,
different linguistic environment and individual differences.
6.2 The data and participants
6.2.1 The intermediate corpus
To my knowledge, there is no freely available corpus of intermediate Norwegian learners of
English. This meant that I had to create my own corpus. Junior high school students seemed
an ideal representative of intermediate Norwegian learners of English, so 18 winter (2012)
mock exams in English were collected from a junior high school English class in the Oslo
area. The students in this corpus are homogenous (Norwegian native tongue, none have lived
abroad and all are fifteen years old). The texts are responses to national tests distributed to
tenth graders across the country (see appendix 1 for tasks). The corpus consists of both
argumentative and literary texts and are all free production. In preparation for the winter mock
exam, students were given a booklet (called “FOCUS”) with information regarding the
various topics which would be covered on the exam, and could be used as a source. The mock
exam lasted five hours. Students were allowed access to any aids, excluding communicative
devices such as phones and the internet. The students also had access to spell check on their
computer. Altogether the corpus comprises 24,732 words.
6.2.2 The advanced corpus
For Norwegian students of advanced English, the NICLE corpus was optimal. The NICLE
corpus is a Norwegian subcorpus of ICLE (The International Corpus of Learner English).
This corpus was initiated by Sylviane Granger of the Universite Catholique de Louvain in
1990. The idea was to create comparable corpora of English texts by L2 learners of English
with a specific L1 background. The texts for the corpora, produced by first-year university
students, are mostly argumentative essays and about the same length, covering the same
topics. ICLE contains approximately 4.5 million words and has sixteen subcorpora (McEnery
& Hardie 2012: 82).
Intermediate Corpus Advanced CorpusL1 Norwegian NorwegianL2 English EnglishSize of corpus 24,732 words 24, 299 wordsNo. of participants 18 39Age of participants 14-15 years old 19-21 years oldYears of English at school 9.5 years 9-10 yearsGenre Argumentative / Literary ArgumentativeType of texts Mock exam Take-home essayTime limit 5 hours No limitReference tools allowed Yes Yes
49
NICLE was compiled in 1992-2002 by Lynell Chvala and Stig Johansson at the
University of Oslo. The total size of NICLE is around 300 essays and consists of
approximately 212,000 words (Johanssson 2008: 117). Most essays are written at home with
no time limit, with the possibility of using reference tools. The topics are mainly
argumentative. For an overview of these topics, see appendix 2.
It is statistically unwise to compare a 200,000 word corpus with a corpus that barely
spans 24,000 words, and also insurmountable to single-handedly analyze a corpus the size of
NICLE. In order to compare NICLE to the intermediate corpus, the NICLE corpus had to be
decreased in both size and participants, i.e., only a few select texts were chosen for the
advanced corpus used in this study. For optimal comparison, a few filters had to be set when
choosing which NICLE texts would be appropriate for the advanced corpus. Ideally, two
comparable corpora should be the same size and have the same number of participants. This
posed a problem, since the NICLE texts are much shorter. For this study, 39 texts (and thus
participants) were chosen for the advanced corpus, amounting to a total of 24,299 words. The
intermediate students had a certain background and access to tools. In order to match the
conditions, and also ensure that the advanced students were truly advanced and as
homogenous as possible, the texts from NICLE had to fulfill certain criteria. In the end, the
two corpora had the following similarities and differences8:
Table 1. Comparable information about the two corpora used
8 The intermediate and advanced students have nearly had the same number of years of English at school because of the new reform, which allowed the intermediate students to begin English from the first grade. However, this does not make the intermediate students less intermediate or the advanced students less advanced.
50
6.3 Procedure
6.3.1 Locating lexical errors
After assembling the two corpora, careful readings of each corpus were performed to identify
lexical dissonance. Once a lexical dissonance was located, it was labeled for its route, effect
and influence. All types of dissonance were recorded and can be found in appendix 3
(intermediate) and appendix 4 (advanced). Information on each individual type of dissonance
can be found in the appendices, with the number of times each dissonance occurred, its route,
effect, the right L2 item (if one could be discerned) and the possible L1 influential item if
relevant.
The primary judger of lexical errors was me (a bilingual Norwegian/English speaker
with L1 English). Each dissonance was only given one route/effect/influence. If a dissonance
was for instance both transliterational and associational, the dissonance was awarded the route
with the perceivably largest influence. In cases where this was ambiguous or difficult to
judge, my supervisor (Signe Oksefjell Ebeling) aided me in the decision. Occasionally a
dissonance fit none of the categories as described by Hasselgren (for instance my definition of
perceived equivalents). In these cases, I altered the taxonomy to fit the data.
6.3.2 Some terms used in the data analysis
In the quantitative analysis, dissonances are represented in two ways: by tokens and types.
Tokens account for the actual instances of word forms found. For instance, if there are 15
wrong instances of the word rose, this means that 15 instances of rose were found in total. A
type is instances of unique words, i.e. if the word rose is wrongly chosen 15 times, it is still
only one type of dissonance. In order to determine the relationship between tokens and types,
a repetition-ratio is calculated. The repetition-ratio is calculated as follows:
For example, if there are 15 tokens and 5 types found in cognates, the repetition-ratio is 3.
Each type of cognate is repeated on average 3 times in the student texts. The higher the
repetition ratio, the more repeated a mistake is. Repeated mistakes are viewed as more
51
problematic – e.g. if rose has a high repetition-ratio, this means rose is frequently repeated
and problematic to L2 speakers.
6.3.3 Statistical Analysis
This study aims to find out whether the difference in number of intermediate and advanced
dissonances is statistically significant. For example, 97 cognates are found in the intermediate
texts and 46 cognates are found in the advanced texts. By using a statistical test to find p-
values, it can be determined if the difference in number of cognates in these two groups is
significant. A p-value of less than 0.05 indicates that the difference is significant.
Statistical significance is calculated by using a log-likelihood test. Paul Rayson's log-
likelihood calculator9 was used, since it is specifically meant for comparing corpus data. The
number of dissonances found in each category are entered in the calculator (in this case, 97
intermediate and 46 advanced cognates) along with corpus size (24,732 words for the
intermediate corpus and 24,299 words for the advanced corpus). The calculator then provides
an LL-value (log-likelihood value). The p-value is determined based on the critical LL-value.
They correspond in the following way10:
Table 2. LL-values to p-values
9 http://ucrel.lancs.ac.uk/llwizard.html accessed 01.05.2013
10 Chart taken from: http://ucrel.lancs.ac.uk/llwizard.html , accessed 01.05.2013
P-value Critical LL-valueP < 0.05 3.84P < 0.01 6.63P < 0.001 10.83P < 0.0001 15.13
52
No. of tokens No. of types No. of L1 influencesDissonances 1077 476 603
Cores SynonymsTransliterations CognatesAssociations PE
53
7. DATA ANALYSIS
This section will present the findings of an error analysis of two corpora (intermediate and
advanced) based on the taxonomy of lexical dissonance. First a general overview will be
given, with a summary of some of the total statistics for in this study. Second, the two sets of
data (intermediate and advanced) will be contrasted for routes, effects, the route/effect
combination and influence, with discussions of some of the results.
7.1 General Overview
Table 3. Number of total dissonances found in the Norwegian learner data (intermediate and advanced)
From a body of text consisting of 49,031 words, 1077 of the lexical words are dissonant in
some way. Two percent of the entire corpus, or 1 in every 50 words is dissonant (although not
all words in a text are lexical). These 1077 dissonances are represented by 476 different types.
The repetition-ratio is 2.3, meaning each dissonance is on average repeated a little more than
twice. Fifty-six percent of the tokens have a perceivable L1-influence.
The routes and effects are distributed as follows:
Figure 5: Distribution of routes in corpora Figure 6: Distribution of effects in corpora
Styl/Con Semantic CollocationalSyntactic Invalid
54
The routes and effects have the following percentage distribution in descending order:
Figure 7: Percentage distribution of routes and effects in corpora
7.1.1 Routes
Cores is the largest route, indicating that the largest lexical coping mechanism for Norwegian
learners of English is using a non-specific vocabulary (e.g. like instead of enjoy or
appreciate). Cores represent as much as 4 in 10 wrong words.
Synonyms is the second largest route, showing that dissonance in Norwegian learners of
English will often be caused by picking a somewhat related, but wrong word (seemingly for
apparently or going for walking). Every 1 in 5 wrong words is caused by this.
Transliterations is the third largest route, giving evidence of literal translations for a little
more than 1 in every 6 wrong words (get away (få vekk) instead of remove, to make a party
(lage fest) instead of throw a party)
Cognates is the fourth largest route, illustrating that interference of similar words between
two languages causes a wrong word choice for 1 in every 6 dissonances. Examples include to
mean that (å mene at) instead of to believe/be of the opinion, or to have the impression (ha
inntrykket) instead of be under the impression.
Associations is the second smallest route, indicating that every 1 in 14 dissonances is caused
by a confusion of word form (defiantly for definitely, or the ocean raised for the the ocean
rose).
Perceived equivalents is the smallest route, accounting for 1 in every 33 dissonances. This
would suggest that for every 33rd dissonant word, the L2 speaker did not know the word form
or meaning of what s/he intended to say at all, or had completely misunderstood the meaning
55
of the word (To give a withering look (unknown intended meaning), or to evolve something
instead of develop).
7.1.2 Effects
Stylistic/Connotational is the largest effect, suggesting that the most common lack of
knowledge for wrong words chosen is related to register. Stylistic/Connotational dissonance is
evident in over half of all dissonances, and includes stylistically inappropriate (most often
informal) words such as crazy for mentally ill, or get for receive/find/obtain.
Semantic is the second largest effect, implying that for approximately 1 in every 6
dissonances, L2 speakers are not fully aware of the senses of the word they are using.
Examples include: The divisions of a hospital instead of wards, or a wood for a forest.
Collocation is the third largest effect, illustrating that lack of word combination knowledge
can explain 1 in every 6 dissonances. Examples are: Choose the right decision instead of
make, or increase your look instead of enhance.
Syntactic is the second smallest effect, showing that confusion between words in word
families or case relations occurs in 1 in every 20 wrong words. Examples: Get born instead
of be born or a near-dying accident instead of near-death.
Invalid is the smallest effect, suggesting that in slightly less than 1 in every 20 wrong words,
Norwegian speakers of English will either translate a Norwegian concept that does not exist in
English (life-lie from Norwegian livsløgn, which means “illusions that a person may build
their life around”) or somehow create a word that does not exist (consum for consumption).
No. of tokens No. of types No. of L1 influencesIntermediate Students 704 301 403Advanced Students 373 212 200
56
7.1.3 Influences
Figure 8: Distribution of influence in corpora
Interlingual influence (58%) > Intralingual influence (42%)
Strong L1-influence (29%) = Weak L1-influence (29%)
Interlingual influence is the largest type of influence (58%), and can be subdivided into
strong L1-influence and weak L1-influence.
Intralingual influence is the smallest type of influence (42%), suggesting that in slightly less
than half of all dissonances, there is no perceivable trace of influence from the L1.
7.2 Intermediate and Advanced Comparisons
This study seeks not only to find general distributions of lexical dissonance in Norwegian
speakers of English, but also to compare intermediate and advanced students for
similarities/differences. This section will examine some of these similarities and differences in
routes, effect, the route/effect combination and influences. First a general overview of each
corpus and its dissonances is presented:
Table 4. Number of total dissonances found in intermediate versus advanced corpora
57
Table 5. Comparable information of dissonances in the intermediate and advanced corpus
* The percentage of dissonances also includes grammatical words, so in reality, the percentage
of dissonances is much higher
** The repetition-ratio number indicates how many times a dissonance is repeated on average
For the intermediate students, approximately 1 in every 35 words is dissonant, while for the
advanced students, this is 1 in every 67 words. The repetition-ratio is higher in the
intermediate students, indicating that they are more prone to repeating the same type of
mistakes. There is also more perceivable L1-influence in the intermediate corpus.
There is a clear difference between the number of dissonances found in the
intermediate corpus versus the advanced corpus. To see if these two sets of dissonance (704
intermediate tokens and 373 advanced tokens) were statistically significant, a log-likelihood
test was conducted. The log-likelihood for difference in number of dissonances between the
corpora gave an LL-value of 97.63, meaning the difference in number of dissonances was
statistically significant to the 99.99th percentile. This gives a p-value of p < 0.0001. The
intermediate group can therefore be said to have significantly more dissonances than the
advanced group.
7.2.1 Routes
This section will show how the dissonances are distributed across the six routes. This
distribution can be divided into tokens (how many individual dissonances were found per
route) and types (how many different types of dissonances were found per route). There are
therefore two ways of determining which route is largest: through counting tokens (which
route has the most amount of dissonances) and types (which route has the most amount of
different types of dissonances). Additionally, the repetition-ratio can suggest which routes
have the most repeated types of dissonance. In order to determine if the distribution of routes
Intermediate AdvancedSize of corpus 24,732 words 24,299 wordsNo. of participants 18 39Percentage of dissonance* 2.8% 1.5%Repetition-ratio** 2.3 1.8Percentage of L1-influence 57% 54%
CoreSynonym
CognateTransliteration
AssociationPE
0
50
100
150
200
250
300
350
Routes
Intermediate Advanced
No
. of
toke
ns
Route Intermediate Students Advanced StudentsNo. of tokens No. of types No. of tokens No. of types
Cognate 97 45 48 25Transliteration 93 68 74 53Perceived Equivalent 21 14 8 8Synonym 130 102 112 91Association 52 47 21 19Core 311 25 110 16Total 704 301 373 212
58
between intermediate and advanced is statistically significant, a log-likelihood test will be
conducted. Lastly, there will be a comparison of the distribution of routes within the corpora
themselves. A comparison of distribution can illustrate if the proportion of distribution is
different or the same. For instance, the intermediate students have 18 more synonymous
dissonances than the advanced students. However if we look at the proportion of routes within
the intermediate and advanced corpora, synonyms account for 19% of intermediate routes and
30% of advanced routes. This means that although the intermediate students have more
instances of synonyms, they rely less on this route than the advanced students proportionally
speaking.
Here is the total distribution of routes for intermediate and advanced students.
Table 6. Total number of tokens and types in routes for intermediate and advanced corpora
7.2.1.1 Comparison of tokens
Figure 9: Ranking of routes in tokens for intermediate and advanced students
SynonymTransliteration
AssociationCognate
CorePE
0
20
40
60
80
100
120
Routes
Intermediate Students Advanced Students
No
. of
typ
es
59
The token ranking shows that cores constitute by far the most tokens for intermediate
students. Cores are high for advanced students, but are surpassed slightly by synonyms (by
two tokens). The fact that students choose the wrong word (synonyms) just as much as they
inappropriately use basic words suggests that advanced students will more often pick the
wrong word instead of “clinging” to core words. Intermediate students, however, prefer the
safety of core words. If we assume that advanced students have a larger vocabulary, a source
of dissonance could perhaps be the choices that come with coping with a larger vocabulary.
The result will be dissonances from the many choices an advanced L2 speaker faces when
having to choose the “right” word in context.
For both groups synonyms rank high (18% for intermediate and 30% for advanced),
and association and perceived equivalents rank low. In the advanced student texts,
transliterations are preferred to cognates. The preference for cognates and transliterations
appears to be equal for intermediate students.
7.2.1.2 Comparison of types
Figure 10: Ranking of routes in tokens for intermediate and advanced students
A type ranking illustrates that although cores have the most tokens, the most varied type of
dissonance is synonyms. Cores may have the most dissonances in total, but they are often the
same types of dissonance that are repeated several times, thus making cores fall from the
largest token category to the second smallest type category. Synonyms and transliterations
have many more different types; therefore generalizing and categorizing why students make
synonymous and transliterational dissonance is much harder, since the types of dissonances
students make in these categories are so varied. While cognates and transliterations have
Core Cognate PE Transliteration Synonym Association0
2
4
6
8
10
12
14
Routes
Intermediate Advanced
No
. of
time
s ty
pe
is r
ep
ea
ted
60
roughly the same amount of tokens (particularly in the intermediate group), transliteration has
many more types in both groups. This can be caused by several factors: a) There are fewer
cognates than transliterations, b) only certain types of cognates are confusing to Norwegian
speakers of English, while transliteration is generally confusing, c) the English cognates that
confuse Norwegian speakers of English are frequent in the L1 and/or d) transliteration is
viewed as more acceptable and transferable than cognates.
7.2.1.3 The repetition-ratio
Figure 11: Repetition-ratio for routes in intermediate and advanced students
The repetition-ratio shows that cores are indisputably the highest repeated dissonances. Core
dissonances are repeated between 7 (advanced) to 12 (intermediate) times on average. In
tokens it was suggested that several core items are “replaced” by synonymous dissonance in
the advanced students. An overview of the repetition-ratio reinforces that not only do
advanced students have fewer core dissonances, but the core dissonances found in advanced
texts are also repeated considerably less.
Cognates are the second most repeated route, with each dissonance repeated on
average two times for both groups. This would suggest that the same types of cognates are
confusing to Norwegian speakers of English (e.g. mean - mene). Due to the high repetition-
ratio for cores and cognates, lists of the most common cores and cognates can therefore be
made for Norwegian speakers of English, so they can be aware of their most frequent
dissonances.
Route LL-value P-value Result
Cores 96.47 P < 0.0001 √Cognates 16.03 P < 0.0001 √Associations 13.05 P < 0.001 √
5.81 P < 0.05 √Transliterations 1.84 P > 0.05 xSynonyms 1.04 P > 0.05 x
PEs
61
Perceived equivalents, transliterations, synonyms and associations all have only
around one dissonance each per type, suggesting that students who use these routes do not
tend to repeat the same dissonance. The consequence of this is that creating lists for these
types of dissonance for these four routes is much more challenging, since many of these
dissonances are idiosyncratic (pertaining to the individual L2 speaker).
7.2.1.4 Statistical significance
There are clear differences in the number of dissonances found in routes for intermediate and
advanced students. In order to find out if the difference between these two groups was
statistically significant, a log-likelihood test was conducted on each route:
Table 7. LL-values and p-values for statistical differences between intermediate and advanced routes
√ = difference between intermediate and advanced is statistically significant
x = difference between intermediate and advanced is not statistically significant
The results of the log-likelihood test suggest that there is a statistically significant difference
in cores, cognates, associations and perceived equivalents between the intermediate and
advanced student texts. There are far more dissonances for the intermediate students for these
routes. Particularly “cores” gives a high LL-value, showing that this route is by far the largest
divergence between the two groups. However, the differences in transliterations and
synonyms in the two groups are not statistically significant, indicating that advanced students
have enough dissonances in these two routes for there not to be a significant difference
between the two groups.
62
7.2.1.5 Comparative distributions
Figure 12: Comparative distribution of routes in
intermediate students
Figure 13: Comparative distribution of routes in
intermediate students
Intermediate students
Cognate TransliterationPE SynonymAssociation Core
Advanced students
Figure 14: Comparative distribution of routes for intermediate and advanced students
The comparative distribution confirms that, proportionally speaking, intermediate and
advanced students have different ways of choosing the wrong word. The difference between
distribution of cores within the two groups is 15%, confirmed by the log-likelihood test to be
the most statistically different way of choosing the wrong word. Proportionally speaking, the
advanced students also show a preference for synonyms and transliteration compared to the
intermediate.
63
7.2.1.6 Summary of routes
Cores are the preferred route for intermediate students, while this preference is distributed
among cores, synonyms and transliterations among the advanced students. This is reflected
both by the log-likelihood test, which showed cores to be the highest LL-value and synonyms
and transliterations to not be significantly different, and the comparative distribution, which
has different percentage distributions for the three routes. Not only are there fewer core tokens
for the advanced students, but core dissonances are also less repeated by advanced students,
as illustrated by the low ranking in the repetition-ratio. The growing preference for synonyms
in advanced students may be explained by the fact that a larger vocabulary can be confusing
to an advanced student, and the trouble of the L2 lies less in clinging to “known” words and
more in coping with the copious amounts of choices an advanced student faces when picking
the “right” word. The differences between intermediate and advanced routes suggest that the
more vocabulary a student has, the less s/he will rely on cores. However, this larger
vocabulary that comes with being at an advanced level is still confusing, as it constitutes more
choices for the L2 speaker. Synonymous dissonance generally has more types than cores,
indicating that core dissonances can be more easily “fixed” in the EFL classroom by
examining core word mistakes that are often repeated.
When it comes to cognates and transliterations, they are preferred equally by the
intermediate students, while transliterations are preferred by advanced speakers. This suggests
that advanced students prefer to “translate” Norwegian phrases, word combinations and
stylistic elements, but why this is a preferred route among more advanced speakers is not
clear. Cognate dissonances have a higher repetition-ratio, meaning they are generally repeated
more often than transliterational ones. This suggests that a) certain cognates are more
frequently misunderstood by Norwegian speakers of English, b) the cognates which are
misunderstood are vital to the structure of the L2 and/or c) transliterations are seen as more
acceptable.
The routes “association” and “perceived equivalents” are the smallest token groups for
both intermediate and advanced students. However, “association” has a high ranking in types.
The reason for this is that these mistakes are highly individual, so the same dissonance is
almost never repeated. The same can be said of perceived equivalents. These two routes are
both small and difficult to categorize, and proved to be the least problematic for students at
both an intermediate and advanced level.
The log-likelihood test confirms a statistically significant difference between all routes
Stylistic/ConnotationSemantic
CollocationSyntactic
Invalid
0100200300400500
Effects
Intermediate Advanced
No
. of
toke
ns
Effect Intermediate Students Advanced StudentsNo. of tokens No. of types No. of tokens No. of types
Semantic 117 72 51 40Collocation 88 69 54 40Style/Connotation 432 101 240 106Syntactic 36 30 21 19Invalid 31 29 7 7Total 704 301 373 212
64
except transliterations and synonyms. This indicates that the other routes are used much more
by the intermediate students, while transliterations and synonyms are popular among the
advanced students as well. The comparative distribution shows that routes have different
proportional distributions within the two groups, i.e. not only are there many more
dissonances in the intermediate texts, but the way they are distributed across routes is
proportionally different.
7.2.2 Effects
This section will show how the dissonances are distributed across the six effects. This
distribution can also be divided into tokens (how many individual dissonances were found per
effect) and types (how many different types of dissonances were found per effect). Effects
will be treated the same as routes: first the token rank will be presented, then the type rank,
repetition ratio, log-likelihood test for statistical significance and lastly the comparative
distribution.
Table 8. Total number of tokens and types in effects for intermediate versus advanced students
7.2.2.1 Comparison of tokens
Figure 15: Ranking of effects in tokens for intermediate and advanced students
Style/Connotation Semantic Collocation Syntactic Invalid0
20
40
60
80
100
120
Types
Intermediate Students Advanced Students
No
. of
typ
es
65
Stylistic/connotational dissonance constitutes by far the largest category in both intermediate
and advanced students. The distribution of stylistic versus semantic effects suggests that
Norwegian students of English struggle more with how words are stylistically marked than
their actual meanings. This does not mean that semantic dissonances are non-existent – they
are somewhere between 14% (advanced) and 16% (intermediate) of all dissonances. In
addition, as mentioned in section 6.1.2, a weakness with free production texts is that there is
no access to the speaker's intended meaning – i.e., if a particular word works in a certain
context, there is no way of knowing if the student was thinking of another item or concept (for
example, if a students writes she broke a cup but is thinking of a vase). So it must be noted
that this effect could be much higher, and would probably be much more evident if translation
texts had been examined, as confirmed by Hasselgren, who found semantic to be the largest
effect.
The third largest effect is collocational dissonance for intermediate students, while it is
tied with semantic dissonance for advanced students, indicating that advanced students
struggle equally with meaning and the combination of different words. Syntactic and invalid
are small in both groups – syntactic dissonance is between 5% (intermediate) and 6%
(advanced), while invalid is between 2% (advanced) and 4% (intermediate).
7.2.2.2 Comparison of types
Figure 16: Ranking of effects in types for intermediate and advanced students
Style/Connotation Semantic Collocation Syntactic Invalid0
0,5
1
1,5
2
2,5
3
3,5
4
4,5
Repetition-ratio
Intermediate Advanced
No
. of
time
s ty
pe
is r
ep
ea
ted
66
Types inform us that the categories hold the same rank in both tokens and types. However,
particularly stylistic/connotational dissonances have many more different types in the
advanced students. Collocation also shows a smaller gap between intermediate and advanced
students in types, suggesting that stylistic/connotational and collocational dissonances in
advanced students are more varied – there are for instance over 100 different types of stylistic
dissonances in both groups.
7.2.2.3 The repetition-ratio
Figure 17: Repetition-ratio of effects for intermediate and advanced students
The repetition-ratio suggests that stylistic/connotational dissonances are the highest repeated
dissonances in both intermediate and advanced students. A reason for this could be the
inclusion of lexical teddy bears – words that are repeated several times and are stylistically
inappropriate are labeled as “cores,” even though they may also be transliterational or
synonymous in nature (see section 5.2.1.6.4). Semantic and collocational dissonances also re-
occur, as for instance each semantic dissonance occurs on average 1.5 times for intermediate
students.
7.2.2.4 Statistical significance
The log-likelihood test showed which of the effects were statistically significant.
Effect LL-value P-value Result
53.26 P < 0.0001 √Semantic 25.49 P < 0.0001 √Invalid 21.91 P < 0.0001 √Collocation 7.63 P < 0.01 √Syntactic 3.73 P > 0.05 x
Styl/Con
67
Table 9. LL-values and p-values for statistical differences between intermediate and advanced effects
√ = difference between intermediate and advanced is statistically significant
x = difference between intermediate and advanced is not statistically significant
The results of the log-likelihood test suggest that there is a statistically significant difference
in all effects but syntactic. Moreover, advanced students are shown to struggle much less with
stylistic and semantic dissonance than the intermediate students, whereas the gap between
advanced/intermediate dissonance in the collocational and syntactic categories is much
smaller (i.e. both groups struggle with these two effects). However, while collocational
dissonance is still found to produce a statistically significant result, this is not the case for
syntactic dissonance.
7.2.2.5 Comparative distribution
Figure 18: Comparative distribution of effects for
intermediate students
Figure 19: Comparative distribution of effects for
advanced students
Intermediate
Semantic CollocationStyle/Connotation SyntacticInvalid
Advanced Students
68
Figure 20: Comparative distribution of effects for intermediate advanced students
The comparative distribution gives some interesting insight: it appears that although each
effect is larger in intermediate students, the distribution within each corpus is almost identical
in the two groups, i.e. the advanced distribution of effects is only proportionally smaller. This
would suggest that although intermediate and advanced students have a different distribution
of routes, or ways of choosing the wrong word (see figures 12 and 13), the incorrect
hypotheses they create about these wrong words (or the lack of knowledge they have of the
wrong word) is the same. This may suggest, in concordance with Schmitt's word knowledge
continuum (see figures 1, 2 and 3), that some aspects of word knowledge are more difficult
than others, e.g. that stylistic knowledge is hard to acquire regardless of level of English.
7.2.2.6 Summary of effects
Stylistic/connotational dissonance is unequivocally the largest effect in both groups,
constituting well over half of all dissonances. This suggests that above all, Norwegian
students of English are unaware of what types of registers English words belong to. The
repetition-ratio illustrates that stylistic dissonances are also the ones that are most repeated.
Semantic/collocational dissonance also proves to be problematic for Norwegian
speakers of English, both in terms of types and tokens. Unlike the syntactic and invalid
categories, semantic, collocational and stylistic dissonances are repeated up to several times,
and these dissonances are reflected in several student texts. Syntactic and invalid dissonances
are more idiosyncratic (they depend on how the individual L2 speaker perceives their L2 and
cannot be easily categorized) and are the smallest effects for both intermediate and advanced
students.
Intermediate StudentsRoute Semantic Collocational Syntactic Invalid TotalCognates 42 22 17 5 11 97Transliteration 15 22 43 3 10 93PE 19 2 21Synonyms 24 35 62 7 2 130Associations 16 9 21 6 52Cores 1 310 311Total 117 88 432 36 31 704
Styl/Con
Advanced StudentsRoute Semantic Collocational Syntactic Invalid TotalCognates 12 12 18 5 1 48Transliteration 8 15 45 3 3 74PE 8 8Synonyms 13 25 68 5 1 112Associations 9 2 8 2 21Cores 1 109 110Total 51 54 240 21 7 373
Styl/Con
69
The log-likelihood test gives a significant difference in all effects but syntactic. The
comparative distribution illustrates that although each group (besides syntactic) is
significantly larger for the intermediate students, the proportion of effects is distributed
equally in both groups. This suggests that when L2 students use words they only have partial
knowledge of, the way in which these words are wrong are the same for both the intermediate
and advanced learners.
7.2.3 The route/effect relationship
All routes must have an effect, or a reason why they are dissonant. Closer inspection of
routes and effects will inevitably lead to the conclusion that there is a relationship between the
two: some effects will combine more often with certain routes. This relationship is of interest
because it can give insight into how effects are distributed in routes, i.e., what types of
incorrect hypotheses are formed for which wrongly chosen word? For example, are cognates
mostly wrong because there is semantic, collocational or stylistic dissonance?
Table 10. The route/effect combination of all dissonances for intermediate students
Table 11. The route/effect combination of all dissonances for advanced students
= no existing route/effect combination for the particular corpus
70
7.2.3.1 Distribution of effects in routes
By examining how effects are distributed in each individual route, it is possible to see which
effects are proportionally larger in each route. This section will illustrate the distribution of
effects in routes through stacked bar charts, in addition to providing possible explanations for
why some effects are more reoccurring than others. The routes “cores” and “synonyms” are
much larger than the other routes, and therefore difficult to make the comparison in one single
chart, so the distribution will be divided into the following: cognates, transliteration,
perceived equivalents and associations in one chart, and cores and synonyms in another.
Figure 21: Distribution of effects in routes for
intermediate students
Figure 22: Distribution of effects in routes for advanced
students
CognateTransliteration
PEAssociation
0
20
40
60
80
100
120
Intermediate Students
Semantic Collocation Styl/ConSyntactic Invalid
CognateTransliteration
PEAssociation
01020304050607080
Advanced StudentsN
o o
f to
ken
s
All effects are present in each route, apart from perceived equivalents (and style/connotation
in association). Semantic dissonance is distributed approximately equally in all four routes
apart from intermediate cognates, which show a higher degree of semantic dissonance. This
then indicates that intermediate students are more prone to using cognates when they do not
know the sense of the English word. This is a case for the semantic equivalence hypothesis,
i.e. that the learner assumes the L1 and the L2 item are the same, without actually knowing
how to use the L2 word (see section 4.2.1.1).
Collocational dissonance is discernible only in cognates, transliterations and
associations. This is due to the semantic nature of perceived equivalents – if the student does
not know the word they are writing, how can they possibly combine it with other words?
71
Cognates and transliterations are common routes for tackling lack of collocational knowledge
for intermediate students, that is, it conforms to the word equivalence hypothesis, where
learners will rely on their knowledge of L1 word combinations and apply this to L2 words
(see section 4.2.1.1).
Stylistic dissonance is found only in the L1-based routes (cognates and
transliterations). It would be rare to find two L2 items that are similar to each other both in
word form and meaning while being stylistically different, so therefore no stylistically
associational dissonances could be found. Cognates have less stylistic dissonance than
transliterations. This could also be due to the fact that it is rarer to find an L1 and L2 item that
are similar in word form and meaning, yet differ stylistically than an L1 and L2 item that are
not similar in word form while being similar in meaning and differ stylistically. This could be
explained by the fact that cognates are a narrower route (i.e. that there are fewer cognates than
possible transliterations. See section 8.1 for more info). However, this could also be because
students at both an intermediate and advanced level may simply prefer transferring stylistic
knowledge between words that are similar in meaning and not form.
Syntactic dissonance occurs in cognates, transliterations and associations, and is
highest for associations. This is because word family dissonances are taken into account, i.e.
if a sentence has the right base form of a word, but it belongs to the wrong word class ( love is
difference instead of different). Such dissonances are not L1-based and therefore labeled as
associational. These sorts of associational dissonances are the largest source of syntactic
dissonance. It is also possible for Norwegian speakers of English to be syntactically confused
by cognates and transliteration. Example: to describe [something] very good (forklare [noe]
veldig godt). Such instances are not prominent in cognates (5 tokens for both intermediate and
advanced students) or transliterations (3 tokens for both intermediate and advanced students),
but do exist.
Invalid dissonance is found in cognates, transliterations and associations, and is
trademark of the intermediate students. This could be because a) intermediate students may
not be aware that several concepts only exist in Norwegian, such as å gi et vink (to give a
wink) are not transferable to English and b) intermediate students are more likely to create
nonsensical words (e.g. burning damage).
72
Figure 23: Distribution of effects in synonyms and
cores for intermediate students
Figure 24: Distribution of effects in synonyms and
cores for advanced students
Synonym Core0
100
200
300
400
Intermediate students
Semantic Collocation Styl/ConSyntactic Invalid
No
. of
toke
ns
Synonym Core0
20406080
100120
Advanced students
Cores appear to be almost exclusively stylistically dissonant, with the exception of 1 semantic
token per corpus. This is due to the nature of cores – students will seldom misunderstand the
meaning of good or nice, and grouping these words with other words will rarely lead to
collocational dissonance (a good/nice time). The reason why such words are dissonant is
because they are not appropriate in a formal register. The result is that they are stylistically
dissonant. Synonyms, on the other hand, can be misunderstood on several levels.
Style/connotation is also the largest effect in this route, and the reason for this is the use of
stylistically informal words which are not core words because they are not neutral and
normally have specific connotations, such as crazy person (mentally ill) or old people (senior
citizens). Such stylistic dissonance accounts for between 48% (intermediate) to 61%
(advanced) of all synonyms. Since these stylistic dissonances are not from an L1-based route,
“synonyms” suggests that Norwegian speakers of English use stylistically informal words
they have acquired outside of their L1, i.e. that they are exposed to informal English, and/or
feel more comfortable using a particular word. Synonyms can also be misunderstood
semantically (essential cause (main cause)) and collocationally. This indicates that Norwegian
speakers of English will not always rely on their L1 for meaning and to create word
combinations.
Synonyms and cores are L2-based routes since they rarely have direct influence from
the L1. This means that the effects in cores and synonyms reflect the difficulties of coping
with an L2 vocabulary. Some of these difficulties could be the choices that come with having
to pick words from an incomplete mental lexicon, confusing synonyms that are often linked
73
semantically (essential - main) and the reluctance to use more specific words that would be
more appropriate in context instead of basic words, as in the case of cores and lexical teddy
bears (for instances overusing boring instead of using more specific and marked words such
as dull/ mundane/ meaningless).
7.2.3.2 Summary of effects in routes
All routes appear to have semantic misunderstandings, apart from cores, since this route rarely
includes words that can be misunderstood. Lack of semantic knowledge is remedied mainly
by cognates in the intermediate students, and also transliterations and synonyms to some
extent. The distribution of semantic effects in routes is more even among the advanced
students, suggesting there is no distinct preference for L1-based routes when coping with gaps
in semantic knowledge at an advanced level.
Collocational dissonance is found in cognates, transliterations, associations and
synonyms. Synonyms have the largest amount of collocational dissonance (particularly in the
advanced students), indicating that students will often choose the wrong synonym, with no
clear preference or pattern. However, if we combine cognate and transliterational
collocational dissonance, they will account for exactly 50% of collocations in both groups,
showing that transfer of collocational patterns from the L1 occurs half of the time.
Syntactic dissonance is found in all routes except perceived equivalents and cores. The
most syntactic confusion derives from lack of suffix knowledge of base forms of words
(associational dissonance), but a small percentage of syntactic dissonance stems from L1-
interference, manifested in cognates and transliterations. Invalid, the smallest effect, is found
in cognates, transliterations, synonyms and associations. Invalid dissonance can be divided
into L1-based (concepts or phrases from the L1 that do not exist in the L2) and L2-based
(non-words that are a result of lack of knowledge of spelling (associations) or compound
words that do not have direct influence from the L1.
Intermediate students Advanced studentsStrong L1-influence 190 122Weak L1-influence 244 78
270 173Total 704 373Intralingual influence
74
7.2.4 Influence
The subject of strong L1-influence has been discernible through the L1-based routes thus far
in the data analysis. However, this does not explain interlingual and intralingual influence in
the L2-based routes (L2-based routes can have weak L1-influence). No comparison has yet
been done of how influences are distributed throughout effects either. As seen in table 5,
possible L1-influence can be detected in 57% of the intermediate dissonances and 54% of the
advanced dissonances in total. By examining the distribution, it may be possible to determine
under what circumstances interlingual or intralingual influence is dominant. An investigation
of distribution may indicate how/when students use and rely on their L1, and how/when they
do not. In instances where the L1 is not a factor, suggestions for intralingual interference will
be provided. This section will explore the distribution of inter- and intralingual influences in
routes and effects. This will be done by showing distribution of influence in L2-based routes
and effects, a log-likelihood test to determine if the differences between strong and weak L1-
influence and intralingual influence are statistically significant between the two groups, and a
comparative distribution to see if the proportion of distribution is the same or different
between the two groups.
7.2.4.1 Routes
L1-influence in routes suggests to what extent Norwegian speakers of English depend on their
L1 when they use words they have partial knowledge of. Table 12 shows how many inter- and
intralingual influences were detected in routes. As explained in section 5.2.3, influences are
subdivided into three types, where strong L1-influence represents L1-based routes (cognates
and transliterations), and weak L1-influence and intralingual influence are both found in L2-
based routes (perceived equivalents, synonyms, associations and cores):
Table 12. Distribution of influences in intermediate and advanced dissonances
Table 12 suggests that L1-influences (both weak and strong) account for between 54-62% of
all dissonances. Strong L1-influences account for between 27-32% of dissonances, meaning
students depend directly on their L1 only around 30% of the time when choosing the wrong
word.
Route Intermediate Students Advanced StudentsInterlingual Interlingual
PE 0 21 0 8Synonym 13 117 27 112Association 1 51 1 21Core 230 91 50 110
Intralingunal Intralingual
75
7.2.4.1.1 Distribution of inter- and intralingual influence in L2-based routes
One phenomenon which table 12 does not explain is the distribution of inter- and intralingual
influence in the L2-based routes. Whereas cognates and transliterations have 100% strong L1-
influence, the inter/intralingual relationship is not apparent in perceived equivalents,
synonyms, associations and cores.
Table 13. Inter- and intralingual influence in the L2-based routes
Figure 25: Distribution of influence in L2-based routes
for intermediate students
Figure 26: Distribution of influence in L2-based routes
for intermediate students
PESynonym
AssociationCore
0
100
200
300
400
Intermediate Students
L1 influence L2 influence
No
. of
toke
ns
PESynonym
AssociationCore
0
50
100
150
200
Advanced students
As shown in table 13 and figures 25 and 26, synonyms are not influenced by L1 to a great
extent. Only 10-24% of synonyms are influenced by finer gridding (see section 4.2.1). Cores
show much more evidence of possible L1-influence, through the way core words in
Norwegian and English are perceived (see section 4.2.1.2). Between 45-74% of core
dissonances indicate possible L1-influence. Reasons for this could be the way core words are
perceived in Norwegian and English, and how the norms for writing formal texts is different
between the two languages.
Influence LL-value P-value ResultWeak L1 86.93 P < 0.0001 √
19.73 P < 0.0001 √Strong L1 13.76 P < 0.001 √Intralingual
76
To conclude, strong L1-influences claim between 27-32% of all dissonances. L2-based
routes show traces of finer gridding and possible stylistic influence, and these weak L1-
influences account for between 21% (advanced) and 35% (intermediate) of all dissonances.
The remaining percentages of dissonance (38- 46%) are words that have possible intralingual
(L2) influence.
7.2.4.1.2 Statistical significance
Influences were tested for statistical significance in strong and weak L1-influence, and
intralingual influence. The log-likelihood calculator gave the following results.
Table 14. LL-values and p-values for statistical differences between intermediate and advanced influences
√ = difference between intermediate and advanced is statistically significant
x = difference between intermediate and advanced is not statistically significant
The results of the log-likelihood test suggest that there is a statistical difference in all types of
influence. Weak L1-influence gives the largest LL-value, suggesting that this is where
intermediate and advanced students differ the most. Intralingual and strong L1-influence have
much lower LL-values, reflecting that the difference in these two categories is smaller
between intermediate and advanced students, yet still significantly different.
77
7.2.4.1.3 Comparative distribution
Figure 27: Comparative distribution of influence in
intermediate students
Figure 28: Comparative distribution of influence in
advanced students
Intermediate students
Strong L1-based influenceWeak L1-based influenceIntralingual influence
Advanced students
The comparative distribution of influences shows that influence is proportionally different in
the two groups. Weak L1 influence is much more preferred by the intermediate students, while
advanced students rely on strong L1 influence and intralingual influence. We will return to
potential reasons for this in the next section.
7.2.4.2 Effects
Examining influence in effects suggests how much Norwegian speakers of English depend on
their L1 when they have insufficient knowledge of an L2 item. An examination of the
distribution of L1-influence may give insight into which aspects of word knowledge
Norwegian speakers of English perceive as being more “transferable” from the L1 to the L2
and which aspects of word knowledge they would rather turn to their L2 for. For instance,
since stylistic dissonances have a high percentage of L1 influence, this may indicate that
Norwegian speakers of English perceive style as being similar between Norwegian and
English, and thus acceptable to transfer. Alternatively, the gap of stylistic knowledge in
Norwegian learners of English is much larger than other aspects of word knowledge, and the
way Norwegian learners of English cope with this gap is by L1-transfer.
Effects Intermediate StudentsStrong L1-influence Weak L1-influence Intralingual influence
Semantic 57 1 59Collocation 44 10 34Styl/Con 60 232 140Syntactic 8 0 28Invalid 21 1 9Total 190 244 270
Effects Advanced StudentsStrong L1-influence Weak L1-influence Intralingual influence
Semantic 20 3 28Collocation 27 8 19Styl/Con 63 67 110Syntactic 8 0 13Invalid 4 0 3Total 122 78 173
78
Table 15. Total number of influences found in effects for intermediate students
Table 16. Total number of influences found in effects for advanced students
7.2.4.1.2 Distribution of influences in effects
Figure 29: Distribution of influence in effects for
intermediate students
Figure 30: Distribution of influence in effects for
advanced students
SemanticCollocation
Styl/ConSyntactic
Invalid
0
100
200
300
400
500
Intermediate students
Strong L1-influence Weak L1-influenceIntralingual influence
No
of
toke
ns
SemanticCollocation
Styl/ConSyntactic
Invalid
050
100150200250300
Advanced students
79
The largest difference in distribution of influences in effects appears to be
stylistic/connotational, where intermediate students seem to rely more on equivalent L1 core
words (big - stor), and advanced students prefer to use an L2 word that a) cannot be explained
by any perceivable L1 influence (e.g. a bunch of), or b) is a directly translatable non-core
word (totally - helt). Stylistic and collocational are the only effects that show weak L1-
influence, meaning that these effects are the result of weak finer gridding and the core word
stylistic differences.
Distribution of semantic dissonance suggests that students will either refer to a directly
translatable word for sense meaning, or there is an unknown intralingual influence that causes
the student to make an incorrect hypothesis of what a word means. This suggests that weak
finer gridding does not occur on a semantic level – perhaps because words that are synonyms
are more likely to have different collocational and stylistic patterns than cover a different
semantic space. Syntactic dissonance is affected much more by intralingual influence,
indicating that confusion of word forms usually stays within the L2.
In both learner groups it is clear that stylistic/connotational dissonance has the most
L1-influence. For the intermediate students, 68% of all stylistic/connotational dissonance
suggest a strong or weak L1 influence, and for the advanced students, this influence is as high
as 54%. This would suggest that the highest number of unsuccessful transfers from
Norwegian to English is stylistic. In other words, Norwegian speakers repeatedly use English
words with a stylistically different Norwegian equivalent (big versus stor, for instance). This
accounts for 63% of all stylistic dissonance for Norwegian speakers of English. The source of
the problem is then that English equivalents of Norwegian words do not have the same
stylistic appropriateness, since Norwegian words that can work on both a formal and an
informal level, can decidedly not work on a formal level in English. Stylistic dissonance is
found in both L1 and L2-based routes. 21% (intermediate) and 48% (advanced) of stylistic
dissonances are L1-based. This indicates that advanced students transfer much more stylistic
knowledge directly from their L1, whereas intermediate students have more indirect L1-
influences, such as the case of the different behavior of core words in Norwegian and English.
One must keep in mind that 37% of all stylistic dissonances are not L1-influenced.
This can be explained by the students' preference for core words and colloquial words
(synonyms). This type of knowledge is L1-independent, meaning that L2 preference for
informal words is an intralingual factor.
The second largest L1-influenced effect is collocation. 61% (intermediate) and 65%
80
(advanced) of all collocations have L1-influence, implying that in over half of the cases, when
a student is uncertain of how to combine words (big demand (high), be a part (play) or plane
accident (crash)), there is a definite L1 word combination equivalent which may have
influenced the speaker's choice of word combination. Nearly all of these collocational L1-
influenced dissonances are strong, accounting for as much as 77% (advanced) to 81%
(intermediate) of all collocational dissonance. The rest of L1-influenced collocations are the
result of weak finer gridding, and only a few examples can be found in synonyms and
associations.
The remaining non-L1-influenced collocations, which are mainly synonyms and
associations, account for as much as 39% (intermediate) and 35% (advanced) of collocational
dissonance, suggesting that in slightly less than half of the cases of word combination
dissonances, there is no apparent evidence of an L1 “crutch.” There are several possible
intralingual reasons why these word combinations are not the result of L1-influence: the L2
speaker may have not acquired the correct synonym, has chosen a synonym s/he is more
comfortable with, used the only synonym s/he knows, or chose the wrong of two synonyms (a
smile crossed her mouth (lips)). Choosing the wrong synonym can be an arbitrary choice or a
confusion of word forms (associational dissonance), such as: when words crash (collide).
The third largest L1-influenced effect is semantic, with between 45% (advanced) and
50% (intermediate) of all semantic dissonance suggesting L1-influence. As in the case of
collocations, most semantic dissonance with L1-influence stems from cognates and
transliteration (57 out of 58 tokens for intermediate, 20 out of 23 tokens for advanced). Cases
where L1-influence is not apparent in semantic misunderstanding include perceived
equivalents, where the student uses a word without knowing its meaning (e.g. give a
withering look), synonyms where semantically similar words confuse the L2 speaker
(divisions of a hospital instead of wards) and associations, where similar word forms that
have unrelated meaning will cause the L2 speaker to choose the wrong word (wrong face
prepositions instead of proportions). Semantic dissonance in perceived equivalents,
synonyms and associational routes are therefore primarily intralingual, illustrating that non-
L1-influenced semantic dissonances stems from confusing the right word with a word that is
semantically similar (synonyms), similar in word form (associations), or has no relation to the
correct word at all (perceived equivalents). These intralingual factors account for roughly half
of the semantic dissonance whether the student is at an intermediate or an advanced level.
81
7.2.4.3 Summary of influences
In conclusion, the three largest L1-influenced effects are style/connotation, collocation and
semantic. Stylistic influence is the result of both L1- and L2-based routes, with more
dissonance in the L2-based routes. Intralingual factors suggest preference for use of more
informal words (cores and synonyms). Collocational and semantic dissonance are primarily
the result of L1-based routes, and therefore show strong L1-influence. Intralingual factors for
collocational dissonance include not having acquired the right synonym, preference for the
wrong synonym, arbitrary guesswork (synonym), and confusion of word forms (associations).
Intralingual factors for semantic dissonance include picking a word that is semantically
similar to the right word (synonym), confusing two words that are similar in form
(association), or unknown factors that are highly individual (perceived equivalents).
The log-likelihood test shows that the difference between the two learners groups is
statistically significant in all three influences; all three influences are significantly larger in
the intermediate group. The comparative distribution informs us that the distribution within
each group is different: intermediate students depend on weak L1-influence, and advanced
students prefer strong L1-influence and intralingual influence.
82
83
8. DISCUSSION
This section will revisit the research questions presented in chapter 2. The discussion will first
look at the first research question: how lexical dissonance is generally distributed in
Norwegian speakers of English. Subquestion a-e will be examined in detail. The second
research question, which compares distribution of intermediate and advanced students, will be
examined afterwards, by answering subquestions a-e.
8.1. The first research question
1. How is lexical dissonance manifested in free production texts written by Norwegian
speakers of English (intermediate and advanced)?
a) How is lexical dissonance distributed across routes and effects in Norwegian intermediate
and advanced learners of English?
Lexical dissonance is classified through both routes and effects. For routes, the distribution in
descending order is: cores, synonyms, transliterations, cognates, associations and perceived
equivalents. For effects, the distribution in descending order is: stylistic/connotational,
semantic, collocational, syntactic and invalid. For a full overview of distribution, see figures
5-7.
b) If some routes and effects are larger than others, what are possible reasons for this?
Routes
Routes are the way a speaker chooses the wrong word, and reflects some of the psychological
processes behind “choosing.” For instance, will a learner prefer to pick a word similar to the
L1 (cognates, transliterations) or a simplified term (cores)? Some routes are clearly larger
than others in this study of lexical dissonance. There seem to be two contributing factors:
i. Some routes are “wider” than others. Some routes cover a larger range of types of
lexical dissonance than others. In the case of cognates and transliterations, there are fewer
words between Norwegian and English that are similar in form and meaning (high - høy) than
similar in meaning and not in form (fin - nice). Therefore, there are more opportunities for
84
transliterational dissonance, since the number of cognates are more limited. The same can be
said of synonyms and associations – there are fewer words that are similar in form and
meaning (increase - enhance) than words that are similar in meaning and not form (seemingly
- apparently). Since synonyms and transliterations are “wider” routes, they rank much higher
in types than cognates and associations.
ii. The route is preferred by the L2 speakers. The allowance for more types of
dissonance in some routes, as explained above, cannot explain why, for instance, there are so
many dissonant cognates. Even though some routes will allow for many different types of
dissonance does not mean that speakers will necessarily use the route. Some routes are simply
preferred to others.
A closer look at each route
Cores is the the most frequently chosen route, representing 4 in 10 wrong words. This
overwhelming preference would suggest that Norwegian speakers of English use core
vocabulary in cases where L1 English speakers would use more precise, low-frequency
words, and this occurs often in argumentative texts. This is seen by Schmitt as being one of
the most common mistakes in L2 learners (Schmitt 2000: 155). Norwegian speakers of
English satisfy their lexical needs by using basic vocabulary instead of more specific,
appropriate terms (using big instead of large, tremendous or incredible), and this will often
lead to production-blocking (there were 20 instances of I think instead of I believe/am of the
opinion).
Synonyms is the second largest route and accounts for 1 in every 5 wrong words. This
route shows that Norwegian speakers of English have a range of words available in their
vocabulary, but their knowledge of these words is partial, and choosing between similar words
is a (seemingly) arbitrary decision. Dissonance then occurs when L2 speakers choose the
wrong synonym, either by believing synonyms are interchangeable, or by choosing the
synonym they know/prefer. Since this is a popular route, particularly among the advanced
students, this would suggest that choosing the right synonym can be a challenge and a large
source of dissonance.
The numerous instances of transliterations and cognates (1 in every 6 wrong words
for both routes) illustrates that many Norwegian speakers of English perceive Norwegian and
85
English as having much more in common than they do in reality. Many of the dissonances
found in these two routes support the word equivalent hypothesis, i.e. that Norwegian
speakers of English habitually transfer all aspects of word knowledge from their L1 to their
L2, assuming that the two will have a 1:1 mapping. The strong relationship between semantic
dissonance and cognates suggest that speakers will often assume that cognates have the same
meaning in the L1 and L2, and the strong relationship between transliterations and stylistic
dissonance shows that speakers perceive translatable words as having the same stylistic
markings.
The low number of associational dissonances (1 in every 14 dissonances) versus the
high number of synonymous dissonances suggests that Norwegian speakers of English do not
rely primarily on phonological routes, i.e. that there is more dependence on the meaning
similarity between words than the sound similarity. The small percentage of perceived
equivalents (1 in every 33 dissonances) suggests that Norwegian speakers of English either a)
are reluctant to use words they are very uncertain of or b) have some idea of the words in their
vocabulary (and will therefore not make perceived equivalence dissonance).
Effects
Effects reflect the “incorrect hypotheses” an L2 speaker creates when s/he decides to use a
word s/he only possesses partial knowledge of, i.e. the assumptions speakers naturally make
when they do not fully know a word. The reason why some effects are larger than others can
be due to:
i. The speaker lacks some aspects of word knowledge more than others. For
instance, stylistic dissonance is very frequent. This could be explained by either a) stylistic
dissonance is not covered sufficiently in the EFL classroom or b) stylistic knowledge is harder
to acquire than other types of word knowledge (Schmitt 2010: 31).
ii. The speaker perceives some aspects of word knowledge as being more important
than others. If speakers perceive stylistic knowledge as being less important, they may be
more willing to risk using words they have no stylistic information about. Conversely, if they
perceive semantic knowledge as being important, they will not risk using a word they have no
semantic information about, causing more stylistic dissonance and less semantic dissonance.
This does not necessarily mean that they have more knowledge of other word aspects, only
86
that they are more willing to be wrong in some aspects and not others.
iii. Some aspects of L1 words are viewed as being more transferable than others. In
the case of effects that have high L1-influence, incorrect hypotheses could be made because
the student makes a false 1:1 mapping between an L1 and L2 item. These are instances of
false word equivalences (see section 4.2.1.1) and may explain why students are more willing
to use words they only have partial knowledge of.
A closer look at each effect
Stylistic/connotational dissonance constitutes the largest effect at over 1 in 2 wrong words.
This effect mostly consists of dissonance between informal words in formal contexts.
Influence distribution in effects suggests that this is a large effect because a) students are often
influenced by equivalent L1 items that are stylistically neutral, meaning Norwegian speakers
of English perceives L1 words as being stylistically transferable and b) Norwegian speakers
of English have acquired many stylistically informal words that have no perceivable L1-
influence. This suggests Norwegian speakers of English acquire English through colloquial
sources, such as television and the internet. Norwegian speakers of English do not produce the
“mishmash of styles,” which Martin claims is “characteristic of non-native production,” –
they stay true to one register (informal), but this seems to be the only register they know
(Martin 1984: 131).
Semantic dissonance is the second largest effect, accounting for 1 in every 6
dissonances. The distribution of influences in effects suggests that in half of the cases, words
are semantically dissonant because of the semantic equivalence hypothesis, i.e. the student
will transfer semantic features of an L1 word to a similar L2 word. However, the other half of
semantic dissonances suggest intralingual influence; the difficulty lies in choosing which
synonym is “correct.” Meaning is often focused on in the EFL classroom, and most would
undoubtably agree that it is one of the more important aspects of word knowledge, if not the
most. Therefore, semantic dissonance can reflect that meaning is complex and difficult to
acquire.
Collocational dissonance is the third largest effect, and also at approximately 1 in
every 6 dissonances. Similar to semantic effect, collocational knowledge is transferred from
the L1 to the L2 in approximately half of the wrongly chosen words. A small proportion of
87
dissonant collocates are caused by weak finer gridding, and the rest are due to intralingual
influence. This could mean that when a Norwegian speaker of English is unsure of how to
combine words (particularly which verb will combine with which object), half of the time
they will refer directly to L1 knowledge, the other half they will choose a synonym they
know/prefer, and in some select cases this will be the wrong of two synonyms for which their
L1 only has one word to express (weak finer gridding). Schmitt suggests that collocational
knowledge is difficult to acquire because it calls for vast amounts of exposure to the language
(Schmitt 2010: 31), so this could explain why students struggle with this aspect.
Syntactic dissonance is the second smallest effect, and only counts for 1 in every 20
dissonances. Mostly it is caused by intralingual influence due to confusion of similar words
within a word family, such as a beautifully day. There are a few select instances where
syntactic elements of an L1 word will be transferred, such as man-dominated from
Norwegian mannsdominert (male-dominated). The many cases of intralingual influence
suggest that syntactic knowledge is perceived as less transferable from the L1 to the L2, as
opposed to the other effects. The low number of syntactic dissonance either suggests that this
aspect of word knowledge is less difficult to acquire, the EFL classroom focuses on syntactic
elements more, or students are hesitant to use words if they are unsure how to place them
syntactically in a sentence.
Invalid dissonance is the smallest effect and is also approximately 1 in every 20
wrong words. It accounts for words or structures that do no exist in the L2, and is largely the
result of creating new words and concepts based on the L1, e.g. psychical damage
(psychological damage). Invalid (4%) is primarily the transfer of Norwegian concepts/words
to English (cold degrees from Norwegian kuldegrader) (although it also covers non-existent
words caused by intralingual influences). Although transfer is common in the other effects,
there were only 38 instances of invalid structures in a body of 50, 000 words. This could
imply two things:
a) Norwegian speakers are aware of which concepts can and cannot be transferred from
Norwegian to English. If this is the case, this would also imply that Norwegians have a high
awareness of the contrast between Norwegian and English concepts and culture, since many
invalid structures (life lie - livsløgn) are a reflection of Norwegian culture and how
Norwegians perceive the world.
b) Transfer is practiced frequently, but since so many concepts and words are similar
between Norwegian and English, it is rare to find instances where a Norwegian structure has
88
no English equivalent. A performance analysis of transfer (calculating how many positive
transfers there are versus negative) may be able to confirm or refute such a statement.
c) How do routes and effects combine? Is there a discernible pattern between how a wrong
word is chosen and why it is dissonant?
It was found that some route/effect combinations are more frequent than others.11 Cores are
nearly all stylistically dissonant, because it is rare for core words to not collocate with
surrounding words (good guy/time/food), not work syntactically or be misunderstood. In other
words, the only way a core word can be dissonant is when native speakers would have used
more precise vocabulary in a formal register.
Cognates, transliterations, synonyms and associations all have a more even spread
of effects. This would suggest that words in these routes can be misunderstood on all levels.
Transliterations and synonyms have more stylistic dissonance, suggesting perhaps that
cognates and associations are relied on less for stylistic information. “Associations” has a
large proportion of syntactic dissonances due to that words from the same word family placed
incorrectly in a sentence (beautiful - beauty) are labeled as associational mistakes. Cognates
and transliterations have a high number of invalid dissonances; this is because “invalid” caters
mostly to structures that exist in Norwegian and not English (e.g. life lie - livsløgn), and these
structures can only be found in L1-based routes.
d) What is the proportion of L1-influenced dissonances (interlingual) vs L2-dissonances
(intralingual)?
In this study, 59% of all dissonances had perceivable L1-influence, suggesting that as much as
41% of dissonances can be explained by intralingual factors. Ellis writes that most learner
errors are intralingual in origin (Ellis 2008: 55). However, the findings from this study suggest
that a majority of lexical errors are interlingual in nature.
11 For a full overview of the distribution, see tables 10 and 11
89
e) Will my data have the same distributional patterns for routes and effects as Hasselgren's?
Route distributions
Hasselgren: synonyms > transliterations > associations > cognates = PE > cores
This study: cores > synonyms > transliterations > cognates > associations > PE
Effect distributions
Hasselgren: semantic > collocational > stylistic/connotational > invalid > syntactic
This study: stylistic/connotational > semantic > collocational > syntactic > invalid
The distributional patterns are very different in both routes and effects. The most striking
difference is cores, which ranks highest in my distribution and lowest in Hasselgren's.
Synonyms and transliterations rank high for both studies, while associations, cognates and
perceived equivalents rank on the lower end of the scale for both studies.
The other large difference is stylistic/connotational, which is by far the largest effect in
this study, and only the third largest effect in Hasselgren's study. Semantic effect is prominent
in both groups, and invalid/syntactic dissonance is low.
There are many reasons for the different results. Primarily the type of data can explain
this, as translation data will elicit different types of dissonance than free production texts.
Another possible explanation is that Hasselgren and I label dissonance differently – I might
have considered more core words as “dissonant”, or she might have viewed these dissonances
as being synonymous. A third possibility is other variables – the participants in our groups are
from different time periods and have had different exposure to English, English education in
Norwegian has changed from 1993-2012, or some results could be due to chance.
90
8.2 The second research question
2. Will the number and distribution of lexical dissonances differ between the
intermediate and advanced students?
a) Is there a difference in number of lexical dissonances found between intermediate and
advanced students?
(See table 4).
The intermediate students have 331 more dissonances than the advanced students, i.e. number
of dissonances diminishes by 47% in the advanced texts. This would confirm that there is a
clear difference in number of lexical dissonances for intermediate and advanced student texts.
b) Is there a difference in distribution of lexical dissonance between routes?
(See table 6 and figures 12-14).
The comparative distributions suggest that there is a difference in distribution of routes
between the intermediate and advanced students, even if we disregard difference in size of the
routes. It appears that cores are the preferred route of intermediate students, while advanced
students' routes are more evenly distributed between synonyms, cores and transliterations. The
routes cognates, perceived equivalents and associations do not show any significant changes
in distribution. These results suggest that advanced students have different ways of choosing
wrong words compared to the intermediate students, and prefer a) different synonyms and b)
structures that only work in the L1 (transliterations) when picking a word, whereas
intermediate students will pick “safe” core words.
c) Is there a difference in distribution of lexical dissonance between effects?
(See table 8 and figures 18-20)
Whereas there is a difference in distribution of routes, there is no perceivable difference in
distribution of effects. This would suggest that although Norwegian speakers of intermediate
91
and advanced English have different ways of choosing the wrong word, the effect of these
wrong words remains the same in both groups.
d) If there are differences between intermediate and advanced students, are these differences
statistically significant?
The answer is both yes and no. The difference between number of dissonances overall is
p < 0.0001, suggesting that there is a very significant difference between the two groups.
When it comes to routes, cores, cognates, associations and perceived equivalents showed
positive for statistical significance, while synonyms and transliterations did not (see table 7).
All effects but syntactic showed positive for statistical significance (see table 9). To conclude,
overall there is a significant difference between the two groups, but the distribution of routes
and effects is not statistically significant in all cases.
e) Will the advanced students show proportionally less L1-influence in lexical dissonance
than the intermediate students, illustrating a “reliance on their ever-increasing knowledge of
the target language” as suggested by Taylor (1975)?
(see table 12 and figures 27-28).
The intermediate students show evidence of L1-influence in 62% of all dissonance, with a
lower proportion of the L1-influences showing strong influence, and a higher proportion of
them showing weak influence. The advanced students show evidence of L1-influence in 54%
of all dissonances, with a higher proportion of the L1-influences showing strong L1-influence,
and a lower proportion of them showing weak influence. One can conclude that
proportionally speaking, advanced students show marginally less L1-influence (8%), but there
is a preference for strong L1-influence. This can be explained by the high number of
transliterations found in the advanced student dissonances and the high number of cores found
in the intermediate student dissonances. The fact that advanced speakers prefer transliterations
proportionally more than intermediate students may suggest that advanced speakers rely more
on the knowledge of their first language when they wish to express themselves in ways they
have not learned in their L2. Intermediate speakers, on the other hand, will stick to “safe” core
words, such as big, or nice that have equivalents in their L1 (stor, snill). However, the
advanced students also show more intralingual influence than intermediate students (44%
92
advanced and 38% intermediate). A possible explanation for this is while intermediate
students again rely on “safe” core words, advanced speakers are more prone to using a wide
array of synonyms that they only have partial knowledge of. This shows signs of “reliance in
knowledge of target language” (Taylor 1975: 88). To conclude, advanced speakers show signs
of relying more heavily on strong L1-influence and intralingual influence, while intermediate
speakers tend to rely on basic words that often have an equivalent in their L1 (big - stor).
Interlingual influence should be measured along a continuum, ranging from direct L1-
influence (how goes it - hvordan går det), to possible instances of L1-influence, such as weak
finer gridding (the ideal family female instead of woman (kvinne)). In certain stages of
interlanguage (beginner, intermediate, advanced), L2 speakers rely on different ends of the
continuum (from strong to weak). Taylor's statement suggests that progressing from a
beginner to an advanced L2 stage will mean a parallel progress on the interlingual influence
continuum from strong to weak (i.e. beginner L2 speakers will rely on strong interlingual
influence, while advanced students will rely on weak L1 influence and gradually intralingual
influence). This, however, is a fallacy, as advanced students show a preference for strong
influence. Conversely, many intermediate students are more willing to rely on core words
(weak L1-influence) and “sacrifice” some of the nuances that come with possible positive
transfer.
93
9. CONCLUSION
This study has investigated the distributional patterns of lexical dissonance in texts written by
Norwegian learners of English. The distributional patterns were determined based on a
taxonomy of lexical errors developed by Angela Hasselgren with Norwegian speakers of
English in mind. The taxonomy was altered to fit free production data and categorized lexical
dissonances based on their source (inter/intralingual influence), how L2 speakers chose the
wrong word (routes) and why the word was wrong (effect).
The main research question was: how is lexical dissonance distributed in texts written
by Norwegian speakers of English? A quantitative analysis determined that for routes the
distribution was: cores (39%), synonyms (22%), transliterations (16%), cognates (13%)
associations (7%) and perceived equivalents (3%). For effects, the distribution was:
stylistic/connotational (62%), semantic (16%), collocational (13%), syntactic (5%) and
invalid (4%). For influences, the distribution was: L1-influence (58%)12 and intralingual
influence (42%). Furthermore, it was observed that positive correlations existed between the
relationship of some of the routes, effects and influences.
The second research question stated: will the number of lexical dissonances and their
distributions differ between intermediate and advanced texts? A log-likelihood test
established statistically significant differences in the number of overall dissonances. These
significant differences indicate that intermediate students have significantly more lexical
dissonances than advanced students, which then implies that lexical dissonance is more
commonly found in intermediate speakers of English. In routes, cores, cognates, associations
and perceived equivalents were significantly different between the intermediate and advanced
students, while transliterations and synonyms were not. In effects, semantic, collocational,
stylistic/connotational and invalid dissonance were significantly different, but syntactic
dissonance was not. In influences, all categories (strong L1-influence, weak L1-influence and
intralingual influence) were significantly different.
Comparative distributions suggested that the way students pick the wrong word
(routes) is different between intermediate and advanced students, but how these words are
wrong (effects) is the same. Influences also suggested that advanced students rely more on
direct L1-knowledge when they have partial knowledge of a word, and intralingual influence
is also a large cause of dissonance. Intermediate students are influenced primarily indirectly
12 Strong L1-influence (29%) and weak L1-influence (29%).
94
by the L1 (weak L- influence), and this is largely due to picking inappropriate L2 words that
are appropriate in the L1.
9.1 Pedagogical implications
Llach stresses that in light of results obtained from an examination of lexical errors, it is the
task of the researcher to “deduce and extract the behavioral and developmental patterns of
lexical errors, and design pedagogical solutions to the lexical problems evidenced by the
analysis of lexical errors” (Llach 2011: 91). This thesis has explained some of the “behavioral
and developmental” patterns of lexical dissonance in Norwegian speakers of English, with
results that strongly indicate that patterns of lexical dissonance exist. It should therefore be
possible to create solutions for lexical dissonance that can be implemented in the Norwegian
EFL classroom.
A problem with designing pedagogical solutions from the findings of such a lexical
error study is that it does not take certain factors into account. This study and taxonomy are
mainly descriptive in nature – they explain the types of lexical dissonance, how they are
chosen (routes) and possible influences, but cannot explain fully why some effects are larger
than others. This is caused by other factors, such as:
i. The priorities and goals of the EFL classroom
ii. The L2 speaker's exposure to English and linguistic environment
iii. Developmental patterns that follow a certain sequence of progression
The goals and focus of the Norwegian EFL classroom may explain some of the results – for
instance, if semantic accurateness is more important than register-training to the Norwegian
educational system, Norwegian students of English will then learn and be more aware of
semantic aspects of English. This can then account for why semantic dissonance is so low and
stylistic dissonance is so high. Additionally, even if a student has, say, more semantic
dissonance, s/he may be more wary of this in the EFL classroom because it is a focus point.
The L2 student may therefore consciously use a dictionary or other resources when given the
chance, reducing the number of naturally occurring dissonances.
The L2 speaker's linguistic environment is also a factor – recent trends in Norwegian
youth television habits suggest a large exposure to colloquial American English, and may
explain why students are mainly aware of informal register, and lack the vocabulary of a
95
formal register. Lastly, developmental patterns of lexical knowledge (which there is little
information on, as linguists still cannot agree on what lexical knowledge fully consists of)
may be the cause of some incorrect hypotheses. In other words, it is possible that stylistic
knowledge “lags” behind in the mental lexicon naturally, and cannot be explicitly taught.
Due to these variables, this study can only account for some of the largest areas of
lexical dissonance, and how these areas could be improved upon. In very brief terms,
dissonance reveals that the following areas of lexis are most problematic for Norwegian
intermediate and advanced speakers of English:
i. Lack of knowledge of formal/specific words
ii. Overuse of basic vocabulary
iii. Lack of stylistic knowledge of words
iv. Dependence on L1 for word knowledge
Points 1, 2 and 3 overlap. Because there is a lack of specific vocabulary, L2 speakers
compensate by overusing their familiar, basic vocabulary, and this basic vocabulary is often
too informal. However, point 3 is also independent – stylistically informal words are only
“core” words between 45-72% of the time. This means that 22-55% of all stylistic
dissonances are not caused by overuse of basic vocabulary, but rather the result of the
learner's exposure to informal, colloquial language.
The obvious solution to point 3 would be to expose Norwegian students of English to
more formal/academic English; implementing such a solution in the EFL classroom is an
entirely different matter. Norwegian students of English surely read formal English in class,
but the amount needed to outweigh the informal English input they receive from television
and internet could simply not be covered in an EFL classroom. Therefore, strategies to elicit
stylistic knowledge and build awareness could be a possible solution. An example would be
for a teacher to write several similar words on the board and have students categorize them
based on their formal/informal features. Training students to be aware of stylistic differences
may aid them into independently categorizing words based on their stylistic markings in the
future. However, this does not solve the problem of lack of formal/specific words (point 1). A
solution could be to boost vocabulary, but what type of specific/formal words do the students
need to acquire? If the topic is “slavery in America,” the students have no problem using
specific words and concepts that are associated with that time period, such as segregation,
96
indentured servants and white power.
The words students tend to overuse are in most cases verbs (for instance, the many
cases of get), adjectives (big, good, bad) and adverbs (really, very). It is difficult to explicitly
teach students how to use more precise words in these cases, and would be perhaps easier to
illustrate which words are “core” and should be avoided in formal writing. For instance, if a
teacher is aware that several students in the case are overfond of using I think, they can offer
the solutions I believe/am of the opinion. The important part is teacher awareness of
stylistically informal words. Appendices 3 and 4 provide lists of core words which could be a
good starting point for basic words that Norwegian students of English generally struggle with
and need alternatives for.
Point 4 poses a problem because transfer is a very common and natural phenomenon,
and does not always lead to dissonance. Therefore, it is not something that can or should be
unlearned. Here one should differentiate between strong and weak L1-influence. The L1-
based routes are perhaps more conscious and obvious instances of transfer, and can be dealt
with in the EFL classroom more easily. Weak L1-influence (as in the case of core words) is
often done subconsciously, and is much more difficult to pinpoint. This discussion will refer
to strong or L1-based influences.
False cognates form a relatively small group and can be easily categorized and
explained. Appendices 3 and 4 provide lists of some of the most common false friends
Norwegian students of English struggle with (but it is by no means exhaustive). If teachers
were to create such a list of false friends, perhaps after observing their own students' texts,
they could then explain some of the main differences between Norwegian and English
cognates to build student awareness that not all similar words have a 1:1 mapping.
Transliterations are trickier. As suggested in section 8.1, transliteration forms a much larger
group of types of dissonance. Students frequently transfer phrases, word combinations and
idioms with and without success. This habit is in some cases viewed as positive because
Norwegian and English are similar enough for some transfers to be successful. If a student is
unsure of a word, transfer is a useful resource. However, automatically assuming that all
Norwegian words and English equivalents have a 1:1 mapping will inevitably lead to many
dissonances with strong L1-influence. Therefore, students should be encouraged to use
transfer only as a last resort strategy – if the student knows another, simpler way of expressing
her/himself, it is better to use a “correct” and simpler L2 item than “risking” a transliterational
failure. The best way to prevent negative transfer in the EFL classroom is to stress that
97
Norwegian and English are not as similar as we may like to believe, and therefore not all
transfers will be successful. This can be illustrated by handing out lists of Norwegian phrases
and collocations that have no or a different English equivalent.
9.2 Suggestions for further research
There is of course much more that could have been done with the data given, but due to time
and space constraints (and a range of other factors), these aspects could not be investigated in
this thesis. This section includes suggestions for further work that could have been done with
the data in this study, e.g. eliciting more information through different/more methods, and
further work that could be done by gathering/using different types of data.
Further investigation on the present data:
• Using performance analysis to examine the data. Performance analysis can be
useful to determine how successful versus unsuccessful the students are, e.g. comparing how
many correct and incorrect collocations are found in the texts. This could give an overall
performance evaluation of what students struggle and are successful with in their lexical
production.
• Using another taxonomy of lexical errors to examine the data. Different
taxonomies reveal different types of errors. For instance, Hemchua & Schmitt (2006) created
a taxonomy with 24 categories of lexical errors for a fine-grained analysis (as cited in Llach
2011: 77). Hemchua and Schmitt's taxonomy also includes form-oriented lexical errors, which
were not examined to a great extent in this study. Using such a taxonomy may reveal different
aspects of lexical error.
• Investigation of word classes in lexical dissonance. Classifying dissonances based
on their word class may illustrate which word classes the students struggle with the most. For
instance, what collocational combinations do students find the hardest? Those with verbs, or
those with nouns?
• Grammatical errors and lexical dissonance comparison. A study of grammatical
errors and lexical errors may indicate if the two correlate. For example, will an L2 speaker
who produces many lexical dissonances also produce many grammatical errors?
98
Further work on additional data:
• Longitudinal study. If the intermediate students wrote similar texts in, for example,
five years, this could show if there have been any patterns of development in lexical
dissonance. For this data it is impossible, since the intermediate students were granted full
anonymity when collecting their texts. However, longitudinal studies of lexical errors are
uncommon, and the field would undoubtedly benefit from from such a study.
• L1 and L2 comparisons. By collecting data written in the students' L1 (Norwegian),
it can be determined if the students who struggle with lexical dissonance in their L2 also
struggle with lexis in their L1. Then a possible link can be drawn between L1 and L2 lexis.
Again, for this particular data set it is not possible, since there is no access to information
concerning either the intermediate or advanced students.
• L1 English control group. By comparing the student texts to similar texts written by
an L1 English control group, it can be determined whether some of these dissonances are
common regardless of whether English is L1 or L2. This is particularly useful for stylistic
dissonance. Students were marked as “dissonant” for using basic, non-specific and informal
vocabulary. However, have L1 English students through ages 15-21 acquired a specific,
formal vocabulary? A comparison of L1 English texts could show if L1 students also rely on a
colloquial, non-specific vocabulary when writing argumentative texts, and determine if this is
generally a problem, or only a characteristic of Norwegian intermediate and advanced
English. The ICLE corpus has similar texts written by English native speakers, and a
comparison can be done between the Norwegian advanced and L1 English speakers.
• A replica analysis for comparable results. This study is the only study to my
knowledge that examines distribution of lexical errors in free production texts written by
Norwegian speakers of English. A replica study would therefore be useful to compare results.
If a replica study had more data (i.e. more texts and participants), the results would be more
robust and could strengthen or counter my results.
99
9.3 Final words
This thesis has hopefully demonstrated the importance of lexical words in the second
language vocabulary. The analysis and results of this study suggest that lexical errors occur
frequently and can impede understanding. By identifying the most frequent errors and
developing systems for them, the second language teacher is better equipped to deal with
lexical issues by knowing a) how and why lexical words are wrong, b) which types of lexical
errors to anticipate and c) how to inform L2 learners so that these errors can be prevented.
This study can serve as a basis for anyone in the future who wishes to examine lexical
dissonance further, whether it be Norwegian L1 or any other L1. The taxonomy has been
altered to fit free production data which can allow for larger quantities of data, e.g. national
tests. The lists of types of lexical dissonances located in appendices 3 and 4 can also be used
by curious English teachers in Norway if they wish to know what types of mistakes are most
common for Norwegian students of English.
100
101
10. REFERENCE LIST
Aitchison, Jean. (2012). Words in the Mind: An Introduction to the Mental Lexicon. UK: John
Wiley & Sons, Inc
Carter, Ronald. (2012). Vocabulary: Applied Linguistic Perspectives. USA & Canada:
Routledge
Carter, Ronald. McCarthy, Michael. (1988). Vocabulary and Language Teaching. NY:
Longman Group UK Limited
Caspi, Tal. Lowie, Wander. (2010). “A Dynamic Perspective on L2 Lexical Development in
Academic English;” in Chacón-Beltràn, Ruben; Abello-Contesse Christián; Torreblanca-
López María del Mar. Insights into Non-native Vocabulary Teaching and Learning: 41-61.
Crystal, David. (2008). A Dictionary of Linguistics and Phonetics. USA; UK: Australia:
Blackwell Publishing.
Cruse, D. A. (1986). Lexical Semantics. Cambridge: Cambridge University Press.
Ellis, Rod. (2008). The Study of Second Language Acquisition. Oxford: Oxford University
Press.
Erman, Britt. Warren, Beatrice. (2000). “The Idiom principle and the open choice principle”
in Text. 20, 20 (1): 29-62.
Fiedler, Sabine. (2007). English Phraseology: A Coursebook. Germany: Narr Francke
Attempto Verlag GmbH + Co. KG.
Granger, Sylvaine. Paquot, Magali. (2008). “Disentangling the phraseological web,” in
Phraseology. An Interdisciplinary Perspective: 27-50.
Hasselgren, Angela. (1993). Right words, Wrong words and Different words: an investigation
into the lexical coping of Norwegian advanced learners of English. Unpublished “hovedfag”
thesis: University of Bergen.
Hasselgren, Angela. (1994). “Lexical teddy bears and advanced learners: a study into the
ways Norwegian students cope with English vocabulary,” in International Journal of Applied
Linguistics 5 (2): 237-260.
102
Hasselgård, Hilde. Johansson, Stig. Lysvåg, Per. (2007). English Grammar: Theory and Use.
Oslo: Universitetsforlaget AS.
Henriksen, Birgit. (1999). “Three dimensions of vocabulary development” in Studies in
Second Language Acquisition 21: 303-317
Jackson, Howard. Amvela, Etienne Ze. (2007). Words, Meaning and Vocabulary. London &
NY: Continuum
James, Carl. (1998). Errors in Language Learning and use: Exploring Error Analysis. NY:
Addison Wesley Longman Limited.
Johansson, Stig. (1978). Studies of Error Gravity: Native reactions to errors produced by
Swedish learners of English. Sweden: Acta Universitatis Gothoburgensis.
Johansson, Stig. (2008). Contrastive analysis and learner language: A corpus-based
approach. Oslo: University of Oslo
Llach, María Pilar Agustín. (2011). Lexical Errors and Accuracy in Foreign Language
Writing. UK; USA; Canada: Multilingual Matters
López, Willelmira Catillejos. (Date of publication not specified) “Error Analysis in a Learner
Corpus. What Are the Learners' Strategies?” accessed:
http://www.um.es/lacell/aelinco/contenido/pdf/45.pdf 24.04.2013
Martin, Marilyn. (1984). “Advanced Vocabulary Teaching: The Problem of Synonyms” in
The Modern Language Journal, 68: 130–137.
McEnery, Tony. Hardie, Andrew. (2012). Corpus Linguistics. UK: Cambridge University
Press.
Meara, Paul. (2009). Connected words: Word associations and second language vocabulary
acquisition. Amersterdam: John Benjamins Publishing Co.
Ortega, Lourdes. (2009). Understanding Second Language Acquisition. UK: Hodder
Education.
Pawley, Andrew. Syder, Frances Hodgetts. (1983). “Two puzzles for linguistic theory: native-
like selection and native-like fluency” in (Red) Richards, S.C. And Schmidt, R. W. Language
103
and Communication: 191-226.
Rayson, Paul. Log-likelihood calculator website. http://ucrel.lancs.ac.uk/llwizard.html
accessed 01.05.2013
Richards, Jack C. (1976). “The role of vocabulary teaching,” in TESOL Quarterly, vol. 10,
no.1: 77-89
Ringbom, Håkan. (1987). The Role of the First Language in Foreign Language Learning.
UK: Multilingual Matters Ltd.
Saeed, John I. (2009). Semantics. UK: Blackwell Publishing.
Schmitt, Norbert. (2000). Vocabulary in Language Teaching. UK: Cambridge University
Press.
Schmitt, Norbert. (2010). “Key Issues in Teaching and Learning Vocabulary,” in Chacón-
Beltràn, Ruben; Abello-Contesse Christián; Torreblanca-López María del Mar. Insights into
Non-native Vocabulary Teaching and Learning: 28-40.
Simensen, Aud Marit. (1998). Teaching a Foreign Language. Bergen: Fagbokforlaget
Vigmostad & Bjørke AS.
Sinclair, John. (1991). Corpus Concordance Collocation. UK: Oxford University Press.
Taylor, Barry. (1975). “The use of overgeneralization and transfer learning strategies by
elementary and intermediate students of ESL” in Language Learning 25: 73-107
Vasiljevic, Zorana. (2010). “Word Associations as a Vocabulary Teaching Strategy in an
Advanced L2 Reading Class,” in Chacón-Beltràn, Ruben; Abello-Contesse Christián;
Torreblanca-López María del Mar. Insights into Non-native Vocabulary Teaching and
Learning: 41-61.
Wood, David. (2001). “In Search of Fluency: What Is It and How Can We Teach It?” in The
Canadian Modern Language Review 57 (4): 573-589.
104
105
APPENDIX 1
National winter mock exams in English, 2012/2013. Distributed to 10th graders in
Norway
106
107
108
109
APPENDIX 2
Overview of questions/topics for the NICLE corpus (Johansson 2008: 117)
• Abortion: right or wrong?
• All armies should consist of entirely professional soldiers: there is no value in a system of military service
• A man/woman's financial reward should be commensurate with their contribution to the society they live in
• Crime does not pay
• Europe
• For and against breakfast
• The role of censorship in Western Society
• Feminists have done more harm to the cause of women than good
• Marriage – an anachronism
• In his novel Animal Farm, George Orwell wrote “All men are equal, but some are more equal than others.” How true is this today?
• In the 19th century Victor Hugo said: “How sad it is to think that Nature is calling out but humanity refuses to pay heed.” Do you think this is still true today?
• Marx once said that religion was the opium of the masses. If he was alive at the end of the 20th century, he would replace religion with television
• The prison system is outdated. No civilized society should punish its criminals: it should rehabilitate them
• Most university degrees are theoretical and do not prepare students for the real world. They are, therefore of very little value.
• Some people say that in our modern world, dominated by science, technology and industrialization, there is no longer a place for Dreaming and Imagination.
110
111
APPENDIX 3 (Intermediate Dissonances)
All dissonances found in the intermediate student texts
Total: 704 tokens, 301 types L1 influence: 434 tokens
Explanation of symbols
Bold – dissonance
“Quotation marks” - (Possible) Norwegian L1 influence
Italics – (Suggested) correct English L2 item (There could be many more alternatives than suggested)
(Parentheses) - Examples from text to illustrate dissonance in context (Examples are largely unabridged, meaning that typos, grammatical mistakes and other features are transcribed faithfully)
[Square brackets] – Words that can also be included in the L1/L1 item, but may not be in all contexts
X - No influence (Norwegian)13 / No equivalent L2 item (English)
Ø - Item is avoided, either by restructuring clause or using other, more appropriate words
C – Core word
ʈ – Lexical teddy bear
++ - If the dissonance could be replaced by a wide variety of other L2 items
Numbers
Numbers before the dissonance tell us how many tokens were found for this one type. If there is no number specified, this means only one token was found.
Numbers after the dissonance inform us the sense for the word. If one word has several senses, they will be numbered chronologically: “1, 2, 3” etc.
Ex. 13 Get 2 (13 tokens for “get,” and this is the second sense listed for “get”)
13 Although something is marked as x from Norwegian to English, it does not mean that the structure does not
exist in Norwegian, simply that it was not an influencing factor in why the speaker chose the wrong word
112
COGNATES97 tokens, 45 types L1 influence: 97 tokens
Cognate/Semantic42 tokens, 13 types L1 influence: 42 tokensConsistence - “konsistens” - consistency(and he could feel the sticky consistence of sweat all over his body)3 Clock - “klokken” - Time/it(He had to hurry up, the clock was almost eleven)2 A cut - “ett kutt” - gash(A man who got a big cut on his left arm)2 History – historie – story (I think that both of the histories tell us a very good story)Hold – “holde” - keep(He held his rifle down)Job- “jobb” - process/act (The job of making your body perfect)2 A land - “et land” - country(Almost no other lands were trading with us)To mean that - “å mene” - to believe/be in the opinion of/think that(I mean that both of them try to save themself from something evil)Meet - “møte” - face(This issue is something we meet in our everyday life)Placed - “Plassert” - Located(The town was Yoleville, placed in the very west of New Jersey)8 To see - “se” - look (She saw me in the eyes/ I saw myself in the mirror)4 To train- “trene” - work out/exercise(Some of them train so they loose allot of weight)2 Way - “vei” - path/route(Others choose another way)
Cognate/Collocational22 tokens, 10 types L1 influence: 22 tokensA baby voice - “en baby stemme” - childish voiceTo come anywhere - “komme noensted/vei” - go anywhereGive an opinion - “gi en mening” - offer an opinionA glad voice - “en glad stemme” - cheerful/happy voice2 To have right/the impression - “ha rett/inntrykket” - be right/be under the impression Little summary - “liten oppsummering” - brief summary2 To say something low - “å si noe lavt” - softly/quietly11 Take surgery/vaccine – ta kirugi/vaksine – have surgery/vaccineTo take on clothes - “ta på klær” - put on clothesSee what time brings - “se hva tiden vil bringe/vise” - see what the future will bring
113
Cognate/Stylistic/Connotational17 tokens, 8 types L1 influence: 17 tokensConnotational2 Cold - “kald” - cool(Luckily the cold air-condition was on)Fantasy character - “fantasifigur” - fictional character
Informal2 Come up with – “komme opp med” – think of(Someone will come up with something)Find – “finne”– discover(No one has found oil)4 Fix – å fikse [på] – alter/change(Fix their appearance)3 Much – “mye” – quite a bit(So much/very much)3 Say – si – state(As said in the quote/the magazine says)So-called – “såkalt” – Ø(And a gruesome sound, or the so called ‘’biip’’ went over the whole room)
Cognate/Syntactic5 tokens, 4 types L1 influence: 5 tokensTo describe very good – veldig godt – very wellThe medicine world – medisinverdenen – medical world/medicinal world2 A sew job – en syjobb – a sewing jobVery much rain – veldig mye regn – a lot of rain
Cognate/Invalid11 tokens, 10 types L1 influence: 11 tokensTo come in on - “komme inn på” - x (get into/are accepted into)(You come in on nurse school)Day to day life person - “dagliglivsperson” - x (everyday/common person)(That isn’t the case for a normal day to day life person)A drastic attempt – “et drastisk forsøk” - xTo give a wink - “gi et vink” - wave2 To hold on to his/her life - “holde på livet” - stay aliveA prettier body - “penere kropp” - xPsychical suffering - “psykiske lidelser” - mental illness/sufferingA sour wind - “sur vind” - xTake [someone's] life - “[noe] tar livet hans/hennes” - x (something/one kills a person) (This [accident] almost took his life)Take oneself through - “å ta seg gjennom” - go through/put oneself through(He is not thinking of the danger that he is often taking him through)
114
TRANSLITERATIONS93 tokens, 68 types L1 influence: 93 tokens
Transliteration/Semantic 15 tokens, 12 types L1 influence: 15 tokensAllow - “å la” - let (She allowed her skinny legs to carry her to the bathroom)The theme appears because … -“dukker opp” - is presentCase – “sak” - issue/subject(I like to give my opinions […] and let people know how I feel about the case)To die early - “dø tidlig” - die young(It then mentions that Henrietta on the other hand, died early)You can only lose life on the different side – “andre siden” - other sideFunny - “morsom” - fun(after a funny day)3 To get away - “få vekk” - remove(Use botox to get away rankles in their face)To grow – å vokse – gained popularity/become more common(Plastic surgery has grown)2 To lay [something] down - “legge noe ned” - put downTo live - “å bo” - stay (We had first planed to live at a hotel)To be locked inside - “låst inne” - trapped insideLook at – “se på” - perceive/view/regard(They look at it as an attack)
Transliteration/Collocation22 tokens, 18 types L1 influence: 22 tokens
A plane accident - “flyulykke” - plane crashBoring sky - “kjedelig himmel” - dull skyTo close the door - “lukke” - shut the doorLife contains -“livet inneholder” - life as a [something] means(I learned more about what the life as a nurse can contain)2 A dark voice - “dyp stemme” - deep voiceThe world has developed into - “utviklet seg” - become (?) / turned intoTo do a good try - “gjøre et godt forsøk” - make a valid attemptTo drive a cab - “å kjøre taxi” - take a cabAn evil circle - “en ond sirkel” - a vicious circleTake a fast look - “ta en fort titt” - Take a quick lookInflict damage - “påføre smerte” - cause damageShe held her ears - “holdt for ørene” clamp her ears4 To make - “lage” - cook food/build a plane/throw a birthday partyTo order a plane - “bestille et fly” - book a flightPositive parts - “positive deler” - positive sides
115
Showed up in her mind - “dukket opp i tankene” - appeared in her mindA car standing in the driveway - “en bil står” - parked in the drivewayThrow light on - “kaste lys på” - shed light on
Transliteration/Stylistic/Connotational43 tokens, 26 types L1 influence: 43 tokens
Connotational3 Failed surgery - “mislykket operasjon” - botched surgeryLarge people - “Store mennesker” - overweight/obese people(Many teen age girls think that the fact that you are perfect like you are, is some Pease of skit that large people says to them selves)
Informal7 Actually - “faktisk” – in fact/Ø(Actually too big)All kinds of – “allslags” – Ø/all types(Remove all kinds of small covers and such)Always – “alltid” – Ø(I have always loved fantasy)Anyway – “uansett” – after all/Ø(It was too late anyway)Play a big role - “spille en stor rolle” - play a major role2 Funny – “morsom” – humorous/entertaining(It was very funny)2 Get 8 – “få” – are permitted/allowed(We get to learn)2 Get the feeling – får følelsen– Ø(You get the feeling that the Queen wants to be perfect)Give an answer – “gi et svar” – reply(You can give me an answer)Go as planned – “gå som planlagt” – Go according to plan(Some of the teenagers begins to cry if not birthday go as planned)Go so well with – “gå så bra med” – is successful ++(You can find thousands of pictures of people that it hastens go so well with)Go wrong – “gå galt” – Ø/fail/be unsuccessful(Can very easily go wrong)A mistake happened - “en feiltagelse hendte” - a mistake occurredImportant for saving lives – viktig for å redde liv – vital for saving lives*It hit [someone] – “det slo [noen]” – make someone realize(But then it hit me, even though)2 Pretty 1 (adverb) – “ganske” – quite(It was pretty good)Something like that – “noe sånt” – Ø(Never do something like that agene)
116
Something to do with – “noe å gjøre med” – Ø/in common/are related to(Both here have something to do with saving)Talk about– “snakke om” – discuss(It`s this I`m going to talk about now)That's right – “det stemmer” – Ø(Well, that’s right)4 The text told - “teksten fortalte” - the text informed2 Totally – “helt” – completely/ vastly different(Totally different method)2 Way – “måte” – manner*(There are many different ways to show love in the same way as it’s easier to)Weird – “rart” – strange*/curious/odd/peculiar(It looked weird)Transliteration/Syntactic3 tokens, 2 types L1 influence: 3 tokensMisinterpreted easy – lett – easily2 Look [tired/angry] at [someone] - “se [trøtt/sint] på noen” - look at someone tiredly/angrily(He looked tired at them)
Transliteration/Invalid10 tokens, 10 types L1 influence: 10 tokensBuilding people - “byggefolk” - construction workersBroken into pieces - “knust i biter” - torn to bitsAn eating table - “et spisebord” - a dining table An eating disturbance - “spiseforstyrrelse” - eating disorder To look empty out the window - “stirre tomt ut vinduet” - xTime ran from her - “tiden løp fra henne” - xA safe bed - “en trygg seng” - xShortly said – “kort sagt” - briefly putA strict look - “strengt blikk” - xA thin body shape - “tynn kroppsform” - x
PERCEIVED EQUIVALENTS21 tokens, 14 types L1 influence: 0 tokens
Perceived Equivalent/Semantic19 tokens, 12 types L1 influence: 0 tokensA better curve – x – unknown intended meaning(Most of the girls that take plastic surgery are mostly focusing on getting bigger lips, a better curve or a facelift)Give me an answer if we want to make this a big deal – x – unknown, possibly “make an agreement”3 To catch - x- to understand(She would catch it like that)3 To evolve - x - to develop
117
(We didn’t have the money to evolve something to get rid of the volcanoes)To expose – x - Unknown intended meaning(A moment of silence exposed itself)A mortal glance – Unknown intended meaning Make someone noble – x - Unknown intended meaningPayback – x – Unknown intended meaning(And get payback for all the training has did to get into that size)2 A Profit - x– Goal/intention(Maybe the profit is to reach popularity)2 Regular - x - usual/commonTo visualize – x – Unknown intended meaning(But also a lot of other things you have to visualize before you are going out on a trip)2 To give a withering look – x – Unknown intended meaning
Perceived Equivalent/Invalid2 tokens, 2 types L1 influence: 0 tokensJob community – x -Work force(So to be happy in the job community you go for, is)She said monotony to herself
SYNONYMS130 tokens, 102 types L1 influence: 13 tokens
Synonym/Semantic24 tokens, 20 types L1 influence: 1 tokenI have been remembering all day since – x – every dayApparently innocent blue eyes – x – seemingly innocentBackside – x – downside(There is a backside though)More than adults can ever know – x – could ever knowWe went down to the coast to get on his boat – x – the shoreThe divisions of a hospital - “avdeling” - wardsThey don't take that much damage –x – can't handle I don't tolerate this – x – I won't tolerateThe environment around him stood still - x - surroundingsI grasped my pen out of my pocket – x– grabbed/took out/pulledIt is not immoral to do some changes or improvements – x– bad/negativeTo name – x – To call(and that includes fixing our body, also named as plastic surgery)I asked if she had time to pick me up to work –x– drive me to work/hitch me a rideHe raised up some binoculars from his bag – x – grabbed/took/pulled out/fetchedThose lives couldn't stand up against the one he had ruined – x – amount toTranslated to - x - interpreted as(This can be translated to the thread he lives on)When the shipment was ready, he warned other lab workers – x – he notified
118
5 A wood – x – a forest(Her uncle Brian owns a wood)I would give you the answer now – x – I will give you
Synonym/Collocational35 tokens, 33 types L1 influence: 9 tokensWhen the plane arrives - “når flyet ankommer” - plane landsThe bottom theme – x – the underlying themeA broken ship – x – a wrecked shipTaking no care – x – taking no heedChoose the right decision – x – make the right decisionA car coming closer – x – pull up closerHow could I know? - “hvordan kunne jeg vite det?” - How should I knowI worked to delete it from the earth - “slette noe fra verden” - erase/eradicate itThe volcano detonated – x – the volcano eruptedSleeping drugs – x – sleeping pills/sedativesOn the edge to die – x – on the verge of dyingEveryone escaped into the mountains - “rømte til fjells” - fled to the mountainsLose a fight - “tape kampen” - lose a battleA job that fits you - “en jobb som passer deg” - suits youGo active – x – become active(Volcanoes went active) Going on a mountain – x – walking on mountainsGo out of bed – x – get out of bed(I went out of bed)Good and downsides – x – upsides and downsidesA happy job – x – A good/satisfactory jobA smile crossed her mouth – x – crossed her lipsGood thing/negative thing – x – good/badIn nice shape – x – in good shapeThe nice side – x – the bright sideI remember it like it was the previous week – “sist uke” – last week2 A quick move - “en rask bevegelse” - A swift moveTo quiz a survey – x – to take a surveyTo reach popularity – x – gain popularityTo squeeze a needle – x – to inject a needleSurrounding that question – x – concerning that question2 Take the warning – x – listen to/heed the warningTell a speech – x – hold a speechThe water almost touched the roof tops - x - reached the roof topsUse a plane – x – take a plane
Synonym/Stylistic/Connotational62 tokens, 43 types L1 influence: 2 tokensConnotational
119
She was too busy [eating] the cadaver – x– too busy [eating] the dead bodyA regular flood – x – an ordinary floodThe magazine should be more correct – x – accurateSlightly horrendous – “litt fælt” – quite/terribly/incredibly
FormalExecute your job the best way possible – x – perform/do your jobImportant for a human to have a job s/he enjoys – x – a person to have a jobEveryone ran to the human being – x – the personInstead of things that are sinister and dismal – x x – negative/bad and depressing/sad
InformalAnother thing – x – Ø(Another thing I think is very good)A ton – x – Ø(Better with no make up, than a ton of it)
2 Basically – x – Ø(Basically, people take plastic surgery)
4 Boobs – x - breasts(Boob job/ big boobs)
2 Boring – x – dull/mundane/meaningless(The world would be a boring place to live)
Fake abs – x - abdominal etching2 Fat – x – overweight/obese(They still look fat)2 Figure out – x – understand/comprehend/Ø(I couldn’t figure out what they had to do with focus on staying alive)Finish up – x – complete(I had to finish up the last five months of junior high)2 Get 5 – x – go/travel(To get from one place to another)2 Get 9 – x – enter/arrive(Their main concern was to get to heaven)Get more into – x – Ø(You will get more into what he think about his life)2 Get to know – x – to become acquainted with/learn(Getting to know people at the same time as executing your job)Go downhill– x – Ø/go poorly/not well(The Norwegian economy had gone downhill)2 Go for – x - pursue(When the people that don’t know what job they should go for)2 Gross – x – unpleasant(Working with injured people might seem gross)A huge lump in my throat - “en stor klump i halsen” - a large lumpGet hurt by a knife – x – get injured by2 Keep [on] – x – continue
120
(Her cells keep living after her tragically early death)Kind of 1– x – type of/sort of(It was a good kind of chocolate)Let it go – x – Ø(I guessed it was a routine question so I let it go)Literally – x – Ø(They tried the put some cells in people who were hurt in accidents, and needed a piece of meat, literally)2 Or something – x – Ø(While she watched Animal Planet or something)Piece of shit – x – Ø(Is some pease of skit that large people says to them selves)Put it– x – express(That is exactly how I would have put it)3 Right – x – correct(This statement is right)Stick to – x – stay/perservere(Stick to that job for the rest of your life)Stuff – x – Ø(It was stuff like)Stupid – x - foolish(Her stupid dad)The good news – x – Ø(Midge goes back to the wood to tell the little people the good news)The whole world – x – Ø(Besides being the most beautiful girl i the hole world)Girls my age is upset because of their body –x – discontent4 Way [too many/few/above] – x – Ø(Way too many plastic surgeries / Way above the volcano)You see – x – Ø(You see, we are going to work)
Synonym/Syntactic7 tokens, 4 types L1 influence: 0 tokensAn airplane accident my friend drove – x – x4 get born/die – x – are born/Ø dieShe fetched a deep sigh – x - xThey like each other very well – x – very much
Synonym/Invalid2 tokens, 2 types L1 influence: 1 tokenNew mown grass - “nyklippet gress” - fresh(ly) cut grassBurning damage - x - burn
121
ASSOCIATIONS52 tokens, 47 types L1 influence: 1 token
Association/Semantic16 tokens, 14 types L1 influence: 0 tokens
Not also –x - not only(Not also have they been very unlucky with the surgeries)To approach – x – to approve(Yes, you can do that. I would approach that)To arrive – x – to appear(A new planet had just arrived very close to the Earth)Defiantly – x – definitelyTwo dissimilar ways - x - two different waysEffect – x - affect(It effects the rest of your life)Existing – x – excitementFront head – x – foreheadHer arms were latent on each side of her – x – lyingLiving – x – leaving( I say “yepp lets go... (shouting) by mom I am living”)Prepositions – x – proportions(Or you got born with the prepositions in your face wrong)Raise – x – rise(The ocean raised)3 To select money – x - to collect moneyTaught – x – thought
Associational/Collocational9 tokens, 8 types L1 influence: 1 token
2 When worlds crash – x – worlds collideConsisting picture – x – containing pictures An exaggerated birthday party – x – extravagant birthday party To increase your look – x – to enhance your lookReduce beauty pressure – x – relieve beauty pressure To search for approval - “lete/jakte etter godkjennelse” - seek approval She let her hand rush through her hair – x – run through her hairThe magazine posted – x – published
Association/Syntactical21 tokens, 20 types L1 influence: 0 tokens
Addict – x - addictiveAn argue – x – an argumentHer hair looked beautifully – x – beautiful
122
Can food – x – canned foodIt was completely chaos – x – complete chaos/completely chaoticTo live a bit complex of day to day life – x – to live a complex/complicated lifeLove is difference – x - differentShe was very dramatically – x - dramaticNear dying – x – near-deathSurgery fails – x – surgery failuresA fiction character – x – fictitiousShe said flirty – x - flirtatiouslyIt was not necessarily – x – necessaryFeeling of perfect – x – feeling of perfection/being perfectA reality character – x – real/based on real lifeSeriously – x – in all seriousness2 To hold a speak – x – a speechFight for survive – x – survivalTechnology upgrades – x – technological upgradesA teleport spell – x – a teleportation spellAssociation/Invalid6 tokens, 5 types L1 influence: 0 tokens
To get beater – x – get better2 Rankles – x – wrinklesHe sward – x – he sworeTo tare through flesh – x - tearWreaked – x – wrecked
CORES AND LEXICAL TEDDY BEARS311 tokens, 25 types L1 influence: 230 tokens
Core/Semantic1 token, 1 type L1 influence: 0 tokensThey have got to explain very well – x – they explained
Core/Stylistic310 tokens, 24 types L1 influence: 230 tokens
7 A little bit [of] – “litt” – a little/slightly/some/Ø ʈ(A little bit entertaining)20 A lot [of]– “mye” – quite a bit/much ʈ(Do it with a lot of care)2 Bad – x – unfortunate C(The only bad thing)28 Big – “stor”– large C/ʈ(Too big)7 Deal with– x – cope with ʈ
123
(It is difficult to deal with situations like death alone)5 Do [anything/everything/something] – “gjøre hva some helst/alt/noe” - Ø/sacrifice ʈ(She is ready to do anything to get prettier)23 Get 1 – x – become/be/Ø C/ʈ(Get thinner/ my vision got blurry/The word got smaller)22 Get 2 – “få” – receive/Ø/find/obtain C/ʈ(Where they get mental help/All they could get)7 Get 3 – “få” – have C/ʈ(You got cancer)20 Good – “god”– wonderful/positive ++ C/ʈ(How good it was)4 Great – x – wonderful/excellent/ ++ C(Great body, great time)5 Huge – “stor” – large/overwhelming ++ʈ(A huge responsibility)8 Kind of 2 – “på en måte” – in a way/can be perceived as/Ø ʈ(Kind of ironic)8 Like – “like” – enjoy/appreciate/Ø ++ C/ʈ(I like the way)6 Look 1 (verb) – “se ut” – appear C/ʈ(To help them look better/everything looks so easy)12 Look 2 (noun) – “utseende” – appearance* C/ʈ(A pretty look) 7 Make – “lage/få” – create/suit better/ causes/Ø C/ʈ(This was made for girls /The virus makes humans die)4 Nice– x – wonderful/lovely ++ C(She looked so nice)17 Pretty 2 (noun)– “pen” – attractive/beautiful C/ʈ(Girls worry about being pretty)13 Really – “virkelig” – quite, truly C/ʈ(I'm really glad)7 Sad – “trist” – tragic/unfortunate C/ʈ(A sad story)20 Thing – x – Ø C/ʈ(The thing about)20 Think – x – find/in the opinion of C/ʈ(I thought the text was good)38 Very – “veldig” – quite/extremely ++ C/ʈ
(Very interesting)
124
125
APPENDIX 4 (Advanced Dissonances)
All dissonances found in the advanced student texts
Total: 373 tokens, 212 types L1 influence: 200 tokens
COGNATES48 tokens, 25 types L1 influence: 48 tokens
Cognate/Semantic12 tokens, 8 types L1 influence: 12 tokensBreak human rights - “bryte menneskerettigheter” - violate human rightsTo can - “å kunne” - know(A chance to show what they can)Learn - “å lære” - to teach(He is not able to learn the children what he knows)To group with - “å gruppere (sammen) med” - associate(We group with those we feel we have most in common with)4 To mean – “å mene [at]” - to believe/think/be in the opinion of(Others mean that a person with a physical or mental defect should be spared from living …)Psychic – psykisk – psychological/mental(When it comes to physics and and/or on the psychic level)2 Shall – “skal” - should(Why shall some people have more money than others?)Way - “vei” - path(Somehow the way from education to work seems)
Cognate/Collocational12 tokens, 6 types L1 influence: 12 tokensGive a punishment – “å gi en straff” - hand out a punishment(The punishment handed out by the authorities)3 Live a life – “leve et liv” – lead a life(Improving the life we actually do live)Much or less – “mye eller lite” - a lot or a little(It is your own responsibility whether you earn much or less)2 On the other side – “på den annen side” - on the other handStrong resources - “sterke ressurser” - rich resources(Those with the strongest resources)4 Take an abortion/decision/job - “ta” - have an abortion/make a decision/accept a job
126
Cognate/Stylistic/Connotational18 tokens, 5 types L1 influence: 18 tokens4 Say –“si” – state (Marx once said that religion was the opium of the masses) 8 Fantasy/fantasize – “fantasi/fantasere” – imagination/imagine/think about(It can be relieving to fantasize and dream about fictive stories)2 So-called - “såkalt” - Ø(In 1787, the so-called Founding Fathers wrote the American constitution)2 Sick - “syk” - mentally ill(An unstable and sick person)2 A thinking robot – “en tenkende robot” - sentient robot(The thinking robot was once a dream)
Cognate/Syntactic5 tokens, 5 types L1 influence: 5 tokensTo argument – “å argumentere” – argue(Some even argument that)Discriminating – “diskriminerende” – discriminatory(They are just as discriminating as they accuse the men for being)Man dominated – “mannsdominert” – male-dominated(So in a man dominated business world)Much – “mye” – a lot/plenty(People have much because of modern scientific instruments)A thinking-process – “en tenkeprosess” – a thought process
Cognate/Invalid1 token, 1 type L1 influence: 1 tokenPsychically – “psykisk” – mentally(Will have more practice both physically and psychically)
TRANSLITERATIONS 74 tokens, 53 types L1 influence: 74 tokens
Transliteration/Semantic 8 tokens, 3 types L1 influence: 8 tokensA sign that everything is in order – “et tegn på at alt er i orden” - x (everything is alright)(Everything is straightened up and you have a sign that everything is in order)Contains – “inneholder” – consists of(A totally different lifestyle, which contains fame, publicity and luxury)6 Look upon/at - “se på” - regard/view/consider(All nations look upon each other as independent/ Sometimes they are looked at as the outsiders)
127
Transliteration/Collocation15 tokens, 12 types L1 influence: 15 tokensBe a part – “være en del av” – play a part(Money is an important part in this game)2 To become eighteen – “å bli atten” - to turn eighteen(When you have become 18-19 years old, you still do not know …)Big demand - “stort krav” - high demand(A society that sets big demands)Cut weeds – “kutte vekk ugress” - hack weeds(You do not want to see the weeds, so you cut them off)Develop an idea – “å utvikle en ide” – to adopt an idea/develop an attitude(Our modern society has developed a very wrong idea) 2 Practical exercises - “praktiske oppgaver” - practical tasks(The universities had practical exercises with grades)Know as your own back pocket – “kjenne som sin egen lomme/lommekjent” - know like the back of your hand(The home guard, for instance, knows their district as their own back pocket)As late as yesterday - “så sent som i går” - as recent as yesterday(I have watched a great deal of television and as late as yesterday I watched an episode)Make a revolt - “gjøre opprør” - Ø revolt (As a result, they did not make any revolts against the capitalists) Use or misuse – bruke eller misbruke – use or abuse(Maybe the child will be used or misused)2 Receive access/knowledge - “få tilgang/kunnskap” - gain access/knowledge(To prevent non-European immigrants to receive access to the country)White-skinned people - “hvithudet mennesker” - light skinned people/Caucasians(White-skinned people tend to fraternize and look down upon colored people)
Transliteration/Stylistic/Connotational45 tokens, 32 types L1 influence: 45 tokens4 Actually - “faktisk” – in fact(Actually too big) A little bit– “litt” – a little/slightly/some/Ø(A little bit entertaining)All kinds of – “allslags” – Ø/all types(Remove all kinds of small covers and such)At least – “i det minste” - Ø(At least we put a lot of effort in) 3 Bad – “dårlig”- poor(ly)*(No matter how bad the conditions you lived under were/ Treat someone bad)2 By the way – “forresten” – Ø(And by the way my room looked like a)Do anything – “gjøre hva some helst” - Ø(She is ready to do anything to get prettier)4 Do a [good/bad] job – “gjøre en [god/dårlig] jobb” -perform well ++
128
(But can all of them do a good job?)Easy way out – “lett utvei” – simple solution(This makes abortion an easy way out)2 [Bad/good] enough – “[god/dårlig] nok”– [in]sufficient(But it is not good enough)Find – “finne” – discover(No one has found oil)3 Get a benefit – “få en fordel” - gain a benefit Get 4 – “få” – develop(There are many people who get bad nerves)Get 8 – “få” – are permitted/allowed(We get to learn)Get back – “få tilbake” – have returned(This part will be really to hard to get back if you should ever regret your actions)Get [someone's] head filled – får hodet fylt – Ø(Housewives […] got their head and ears filled up with speeches)2 Get [something/the most] out of – “få noe utav” – receive benefits(Then it is up to others to […] and try to get something out of it)Get [someone's] way – “få viljen” – something goes according to their wishes/Ø(If they got their way, they would send Norway […] back into the patriarchal stone age)2 Kind of– “slags” – type of/sort of(No mather what kind of a prison system)3 Look – “se” – appear/seems(To help them look better/everything looks so easy)Mess– rot – disorder/chaos(The economic mess is in order)Obviously – “tydeligvis” – Ø(Obviously, they were wrong)2 So many [things] – “så mange [ting]” – several/many [things](There are so many things you can say about marriage)Talk about – “snakke om” – discuss(It`s this I`m going to talk about now)3 Take a look at – “ta en titt på” – look at/examine(I am going to take a look at the unequal financial rewards in society)Take care of – “ta vare på” – support(There are also people who just can't take care of their children)Take [someone] out – “ta noen med ut” -bring someone for(They should take her out for a couple of drinks)Take over - “ta over” - occupied/gained control over (Science technology has taken over large domains)Used to – “vant med” – accustomed to(They have to be given necessary guidelines away from what they are used to)Want nothing to do with – “vil ikke ha noe å gjøre med” – does not want to affiliate with(“One Nation” wants nothing to do with Asians)
129
Transliteration/Syntactic3 tokens, 3 types L1 influence: 3 tokensAway from – “vekk/bort fra” - Ø(They have to be given the necessary guidelines away from what they are used to)The danger is small – “faren er lite for at” – There is a small risk(Even though the danger is small that it will ever happen)Small part - “liten del” - small part of the population/fraction(It is obvious that only a small part can afford this lifestyle)
Transliteration/Invalid 3 tokens, 3 types L1 influence: 3 tokensLife-lie - “livsløgn” (illusions that a person may build their life around, first coined by Henrik Ibsen)- x(An average person does need the life-lie to function)Office rat - “kontorrotte” (slang for a person who spend too much time in the office) – x(Maybe dreams still exist with overworked office rats)Self-decided abortion - “selvbestemt abort” - abortion on demand
PERCEIVED EQUIVALENTS8 tokens, 8 types L1 influence: 0 tokens
Perceived Equivalent/Semantic8 tokens, 8 types L1 influence: 0 tokensIn conformity with – x – unknown intended meaning(In conformity with drugs, smoke and alcohol, television is highly addictive)Division – x – unknown intended meaning, possibly gap in time(A division in time will best justify this matter)Everything falls to pieces – x – unknown intended meaning, possibly falls into place(Now I have stated the facta about anachronism, every thing fall to pieces)Earlier – x – unknown intended meaning, possibly formerAs the earlier classes in society are more or less evened out)Existence – x – unknown intended meaning([Military systems] do resemble each other in more than one field of strategy and existence)Programme form – x – unknown intended meaning, possibly television genre(Programme forms such as situational comedies)(They do resemble each other in more than one field of strategy and existence)Significant – x – unknown intended meaning, possibly independent/original(...Ukraine and Pakistan all have their own significant [military] systems)Stick out – x – unknown intended meaning, possibly strike (He also said that political conditions are transformed in the cultural sphere, so that they do not stick out as improper)
130
SYNONYMS112 tokens, 91 types L1 influence: 27 tokens
Synonym/Semantic13 tokens, 11 types L1 influence: 3 tokensAccording to – x – after/as a result of (?)(According to recent in events in Norway)Come up – x – appear/are produced(I am truly amazed by all the fancy products from fancy industries that come up nowadays)Divisions – x – boundaries(Which spoke to people across class, ethnic, and gender divisions)Earthly – “verdslig” – worldly(Both religious and earthly constrictions)Essential cause – x – main/primary cause(Greed and excess consumption have often been essential causes [for exploiting nature])Force – “styrke” - strength(In the extreme case of war, people will fight with an unknown force far greater than …)Guard – beskytte – protect (Censorhip guards leaders against unwanted criticism)Harmonize – x – equal(They often get a punishment that does not harmonize with the crime they have committed)Matter – x – factor(Age is also a matter which creates inequality)3 Notorious (criminal)– x – minor offender(A notorious criminal has committed several minor crimes and get eventually caught)Think – x – find/in the opinion of(I thought the text was good)
Synonym/Collocational25 tokens, 20 types L1 influence: 7 tokens2 Achieve knowledge – x – gain knowledge(The university as a place to achieve their knowledge)One other aspect of seeing it – x – way of seeing itA better level – x – a higher level(A better level of expertise)Keyboard button – x – keyboard key(Push a few buttons on the keyboard)Call wrath – x – incur wrath(They were afraid they might call God's wrath upon them)Connect bonds – knytte bånd (possible associative influence)– forming bonds(Quality-time spent connecting bonds to your children)2 Do a crime – x – commit a crime3 Female – “kvinne” – woman(The ideal family female have been killed)2 Fight – “kamp” – struggle*(Little has changed since the fight [for women's rights] originated in the 1800s)
131
A great step – x – large step(Would be a great step to take)Happy and conflicting times – x – good and bad times (counts as 2 dissonances)Human beings - “mennesker” - mankind (For the good of human beings)Human life - x - life(Religion has always been an important part of human life)Hunted – x– persecuted(Those with such views [non-religious] were hunted)Keep to the facts – x – stick to the factsTo pinpoint a point – x – to make a point/to pinpoint somethingPut someone straight – x – set someone straight(There is no doubt that the military has put several people straight)Put upon – x – imposed on (They would care less about the exploitation put upon them by the capitalists)To take the wrong signal – x – take something the wrong way(When a person is not rewarded he or she may take the wrong signal)A way of punishment – x – a form of punishment
Synonym/Stylistic/Connotational68 tokens, 54 types L1 influence: 17 tokensStylistically too formal2 Human beings – “mennesker” – people(They need to buy new things to be successful human beings)Lady – “kvinne” – woman(In Muslim countries, ladies are subdued by men)
Stylistically too informalA bunch of – x – Ø(A standing army of professionals can be called out to missions much faster than a bunch of nineteen-year-olds)All over again – x – Ø(The army lets people start all over again)A long time ago – x – Ø(This happened a very long time ago)A whole lot more – x- Ø(Modern mobile phones have everything you need, and a whole lot more)
Bad – x – unfortunate (this case is also stylistically unacceptable in Norwegian, so is not transliteration)(The only bad thing)Bad enough as it is – x – Ø(This world is a bad enough place as it is)2 Basically – x – Ø(Basically, people take plastic surgery)Broke – x – poor/poverty-stricken(To prevent that they are broke when they are “on their own” again)2 Brainwash – “hjernevaske” – indoctrinate
132
(Tend to be more brainwashed and focused on his job)2 Called up into the army - “innkalt til militæret” - drafted in the armyCome along – x – Ø/enter the scene/was introduced ++(Then feminism came along and women all over the world)2 Chance – x – opportunity(You never had the chance to see what you really like)2 Crazy – x – mentally ill(Under special circumstances, like if the mother was crazy)Dark-skinned people - “mørkhudet mennesker” - minorities(Dark-skinned people are maybe the most exposed group)Fancy – x – extravagant/lavish ++(I am truly amazed by all the fancy products from different industries)Hook up to - “hooke opp til” - connect to(You start your working day only by hooking up to the company's server)Get along with– “komme overens med” – be on good terms with(You have to get along with your fellow soldiers)2 Get away with – “slippe unna med” – face no repercussions/Ø(Grown people […] expect to get away with anything)Get 5 – x – go/travel(To get from one place to another) Get 6 – x – to give(What to get Monica for her birthday)Get 7 – x – Ø/achieve(You won't get as far)Get 9 – x – enter(Their main concern was to get to heaven)Get back – x – return(When they got back they dinner on the table and a clean house greeted them)Get home – x – come/arrive at homeEx. When he got home3 Get married – x – marriage(I have written down arguments for getting married)Get the opportunity – x – Ø/be able to(The more average people want more money to get the opportunity to be busier)Get through – x – survive(In order to get through it at all)3 Get rid of– x – remove/discard(A way of getting rid of unwanted elements)Give [someone] up– x – give away/have to part with something/one(Carrying a child for nine months, and then give it up)Go away – x – disappear/vanish(The weeds will never go away)2 Go for – x - pursue(When the people that don’t know what job they should go for)Go get [something] – x – pursue(If men want something, they go get it)
133
Hang out – x – spending time together ++(Phoebe and Rachel hanging out at their favorite coffee bar)Has/have got– x – have (It has got the features of a drug)Keep [someone] down – x – oppress(And besides from making people passive, it also kept them down)Keep up– x – stay at the same level ++(Those who can not [afford expensive things], will not be able to keep up)Look – “utseende” – appearance(A pretty look)Make of oneself– x – become successful ++(Few have any idea of what they want to make of themselves)[Have] off – x – Take/have a vacation(He hadn't had a week off work since he was 20)Old people – x – the elderly/senior citizens(When we grow old most of us go living in an institution for old people)Snatch – x – steal/robbery/shoplift(Perhaps a successful snatch triggered a more serious crime)Something on [television] – “noe på tv” – watch television/Ø(Neighbors and friends often came to watch when there was something on)2 Stuck [somewhere] – x – Ø(You are stuck here because you need a job) Stupid – “dum” - Ø(We are made passive and stupid)The whole world – x – Ø(Know what happens in every soap opera in the whole world)In other terms – x – in other words(In other terms, she fought for women's rights)2 Totally– “helt” – completely/ vastly different(Totally different method)Turn out – x – end/occur/Ø(Things did not turn out exactly as planned)Up to – x – responsibility (It is all up to the government)2 [A] waste [of] time – x – Ø/not profitable/worthwhile(There is not point in dreaming. It's a waste of time)
Whatsoever – x – Ø(Often nothing is done to prepare the felons rehabilitation what so ever)
Connotational Severely – x – greatly (Severely can only work in negative contexts)(The situation has improved severely in recent years)
Synonym/Syntactic5 tokens, 5 types L1 influence: 0 tokensGo to great achievements – x – go to great lengths
134
(You don't have to go to great achievements to break away from it)Advantage – x – get the advantage/in [his/her/their]advantage ++(Man has altered many things in order to achieve advantages)Practically good – x – well, practically speaking/practically Ø(If they can do the job practically good)The statement says – x – statement ØEmpty-minded – x – mindlessly(Or more likely surf empty-minded on the internet)
Synonym/Invalid1 token, 1 type L1 influence: 0 tokensInacceptance – x – intolerance(When it comes to inacceptance)
ASSOCIATIONS21 tokens, 19 types L1 influence: 1 token
Association/Semantic9 tokens, 9 types L1 influence: 0 tokensBomb – x – bombard(Companies bombed them with tempting possebilities)Change – x – chance(They are not given the change to prove the crime was …)Deliverance – x - delivery(Humans do the deliverance, but machines do all the preparations)Descending – x – condescending(Now, people think these descending thoughts [racism] to be dead and gone)(Censorship in the classic sense is almost non-exciting)Imported – x – important(The most imported thing is how)Physics – x – physical(When it comes to physics and/or on the psychic level)Preacher – x – priestPreserve – x – protect(Now, almost every country has a kind of constitution that preserves the inhabitants)Principal – x – principle(The principal of punishing criminals is not outdated)Transfer – x - transportation
Associational/Collocational2 tokens, 2 types L1 influence: 1 tokenBoothes – x – boxes(It puts people into boothes which does not excist in real life)Take an assumption – “å anta” – make an assumption
135
(I may have taken wrong assumptions)
Association/Syntactical8 tokens, 6 types L1 influence: 0 tokens
Dependent – x – depending/are dependent(Because different firms dependent on the type of …)Dismiss – x – be dismissed(When they dismiss, they receive a large amount)Excess – x – excessive(Greed and excess consumption)3 A high degree – x – a higher degree(Why should the only ones with high degrees get the job?)In compare to – x – in comparison toSituation comedy – x – situational comedy
Association/Invalid2 tokens, 2 types L1 influence: 0 tokensConsum – x – consumption(They don't have an enormous consum)Predictment – x – prediction (The predictment of marrying is that you want to be together forever)
CORES AND LEXICAL TEDDY BEARS110 tokens, 16 types L1 influence: 50 tokens
Core/Semantic1 token, 1 type L1 influence: 0 tokensThrough with – x – done with/have completed/finished(When you are through with your service)
Core/Stylistic109 tokens, 15 types L1 influence: 50 tokens
21 A lot [of] – “mye” (not stylistically inappropriate) – quite a bit/much/several ʈ(Do it with a lot of care/ A lot of us today)8 Big – “stor” (not stylistically inappropriate) – large C/ʈ(Big changes/trend)5 Boring – “kjedelig” (not stylistically inappropriate) – dull/mundane/meaningless ʈ(The world would be a boring place to live)14 Deal with – x – cope with/Ø ʈ(It is difficult to deal with situations like death alone)16 Get 1 – x – become/be/Ø C/ʈ
136
(Get better)22 Get 2 – “få” – receive/Ø/find/obtain C/ʈ(Where they get mental help/All they could get)6 Get 3 – “få” – have C/ʈ(Ones who got their homes turned upside-down/Money is all he has got)4 Good – god (not stylistically inappropriate) – wonderful/positive/ ++ C(How good it was)Great – x – wonderful/excellent ++ C(They had a great time) 5 Huge – stor (not stylistically inappropriate) – large/vast/overwhelming ʈ(The difference is huge)Make – lage/(not stylistically inappropriate) – create C(You need an idea before you can make something)Nice – x – wonderful/lovely/kind ++ C(In spite of nice words)11 Really – virkelig – quite, truly C/ʈ(I'm really glad) Right – x – correct C(This statement is right)2 Sad – “trist” (not stylistically inappropriate) – tragic C(I found it surprising and incredibly sad)12 Thing – x – Ø C/ʈ(The thing about/Most important thing)7 Very – “veldig” (not stylistically inappropriate) – Ø C/ʈ(Very interesting)
137
APPENDIX 5
Explanation of numberings for get
GET 1 – x – become/be/Ø
Ex. Get thinner/ my vision got blurry/The word got smaller
GET 2 – få – receive/Ø/find/obtain
Ex. Where they get mental help/All they could get
GET 3 – få – have (DELEXICAL VERB!)
Ex. You got cancer
GET 4 – få – develop
Ex. There are many people who get bad nerves
GET 5 – x – go/travel
Ex. To get from one place to another
GET 6 – å gi – to give
Ex. What to get Monica for her birthday
GET 7 – å komme – Ø/achieve
Ex. You won't get as far
GET 8 – få (not stylistically inappropriate) – are permitted/allowed
Ex. We get to learn
GET 9 – x – enter/arrive
Ex. Their main concern was to get to heaven